From 238c966b5da16982d7ffd5b269d63a3823f8d064 Mon Sep 17 00:00:00 2001
From: theiturhs <96874023+theiturhs@users.noreply.github.com>
Date: Sat, 25 May 2024 13:05:34 +0530
Subject: [PATCH 1/3] Add files via upload

---
 .../Cross Validation Techniques.md            | 28 +++++++
 .../CrossValidation_Techniques.ipynb          |  1 +
 .../Hyper-Parameter Tuning.md                 | 78 +++++++++++++++++++
 .../HyperParameter_Tuning.ipynb               |  1 +
 4 files changed, 108 insertions(+)
 create mode 100644 Brain Tumor MRI Classification/Cross Validation Techniques.md
 create mode 100644 Brain Tumor MRI Classification/CrossValidation_Techniques.ipynb
 create mode 100644 Brain Tumor MRI Classification/Hyper-Parameter Tuning.md
 create mode 100644 Brain Tumor MRI Classification/HyperParameter_Tuning.ipynb

diff --git a/Brain Tumor MRI Classification/Cross Validation Techniques.md b/Brain Tumor MRI Classification/Cross Validation Techniques.md
new file mode 100644
index 00000000..43fd7a08
--- /dev/null
+++ b/Brain Tumor MRI Classification/Cross Validation Techniques.md	
@@ -0,0 +1,28 @@
+# Cross Validation Techniques for Brain Tumor MRI Classification
+____
+
+The dataset on which the cross validation is carried out can be found [on kaggle](https://www.kaggle.com/datasets/theiturhs/brain-tumor-mri-classification-dataset/data). Find the implementation of this [in this notebook]().
+
+### Cross Validation Teachniques carried out are as follows:
+*Different techniques for distributing dataset into training and testing dataset*
+* Hold-Out CV
+* K-Fold Cross Validation
+* Repeated K-Fold
+* Leave-One-Out (LOO)
+* Stratified K-Fold
+
+### Different techniques and their obtained training accuracies¶
+| Technique	| Average Accuracy across all classes |	Pituitary |	No Tumor | Meningioma | Glioma |
+| ---- | ---- | ---- | ---- | ---- | ---- |
+| Hold-Out CV |	0.9380 |	0.9860 |	0.9760 |	0.8210 |	0.9690 |
+| K-Fold CV	| 0.9604	| 0.9858 |	0.9858 |	0.9196 | 	0.9420|
+| Repeated K-Fold CV	| 0.9585 |	0.9504 |	0.9838 |	0.9255 |	0.9703|
+| Stratified K-Fold CV	| 0.9548 |	0.9544	| 0.9838 |	0.9096 |	0.9667|
+
+**Out of these techniques, K-Fold CV technique gives better overall accuracy and class wise accuracy.**
+
+1. Hold Out Technique: Randomly the dataset was distributed into 80% training set and 20% validation set. The aacuracy achieved was 93.80%.
+2. K-Fold Cross Validation: This divides data into k equal-sized folds, trains the model k times, each time using k-1 folds as training data and one fold as validation data. The accuracy achieved was 96.04%.
+3. Repeated K-Fold: This repeats the k-fold CV process multiple times with different random splits of the data. It achieved 95.85% accuracy.
+4. Leave One Out: It is a special case of k-fold CV where k is equal to the number of samples in the dataset. Each sample is used as a validation set once. So this was not implemented. It was computationally expensive, would have required a lot of training time.
+5. Stratified K-Fold: It is like k-fold CV, but ensures that each fold preserves the percentage of samples for each class. It achieved an accuracy of 95.48%.
\ No newline at end of file
diff --git a/Brain Tumor MRI Classification/CrossValidation_Techniques.ipynb b/Brain Tumor MRI Classification/CrossValidation_Techniques.ipynb
new file mode 100644
index 00000000..a0b29332
--- /dev/null
+++ b/Brain Tumor MRI Classification/CrossValidation_Techniques.ipynb	
@@ -0,0 +1 @@
+{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"nvidiaTeslaT4","dataSources":[{"sourceId":8392068,"sourceType":"datasetVersion","datasetId":4992016}],"dockerImageVersionId":30698,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"import warnings\nwarnings.filterwarnings(\"ignore\")","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:13:42.125551Z","iopub.execute_input":"2024-05-23T18:13:42.125904Z","iopub.status.idle":"2024-05-23T18:13:42.137485Z","shell.execute_reply.started":"2024-05-23T18:13:42.125875Z","shell.execute_reply":"2024-05-23T18:13:42.136607Z"},"trusted":true},"execution_count":1,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nimport cv2\nimport matplotlib.pyplot as plt\nfrom sklearn.metrics import accuracy_score","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2024-05-23T18:13:47.825566Z","iopub.execute_input":"2024-05-23T18:13:47.826453Z","iopub.status.idle":"2024-05-23T18:13:49.014840Z","shell.execute_reply.started":"2024-05-23T18:13:47.826423Z","shell.execute_reply":"2024-05-23T18:13:49.013951Z"},"trusted":true},"execution_count":2,"outputs":[]},{"cell_type":"markdown","source":"# Cross Validation Techiques\nDividing training and testing datasets for training purpose\n\n### Different techniques for distributing dataset into training and testing dataset\n1. Hold-Out CV\n2. K-Fold Cross Validation\n3. Repeated K-Fold\n4. Leave-One-Out (LOO)\n5. Stratified K-Fold\n\n#### Note: These techniques are implemented over training dataset only. No augmented dataset is considered to make it less complex and produce the results.\n\n### Different techniques and their obtained training accuracies\n\n<table style=\"width:90%; font-size: 16px;\">\n    <tr>\n        <th rowspan=\"2\">Technique</th>\n        <th rowspan=\"2\">Average Accuracy across all classes</th>\n        <th colspan=\"4\">Per Class Average Accuracy</th>\n    </tr>\n    <tr>\n        <th>Pituitary</th>\n        <th>No Tumor</th>\n        <th>Meningioma</th>\n        <th>Glioma</th>\n    </tr>\n    <tr style='text-align: center;'>\n        <td>Hold-Out CV</td>\n        <td>0.9380</td>\n        <td>0.9860</td>\n        <td>0.9760</td>\n        <td>0.8210</td>\n        <td>0.9690</td>\n    </tr>\n    <tr style='text-align: center;'>\n        <th>K-Fold CV</th>\n        <th>0.9604</th>\n        <th>0.9858</th>\n        <th>0.9858</th>\n        <th>0.9196</th>\n        <th>0.9420</th>\n    </tr>\n    <tr style='text-align: center;'>\n        <td>Repeated K-Fold CV</td>\n        <td>0.9585</td>\n        <td>0.9504</td>\n        <td>0.9838</td>\n        <td>0.9255</td>\n        <td>0.9703</td>\n    </tr>\n    <tr style='text-align: center;'>\n        <td>Stratified K-Fold CV</td>\n        <td>0.9548</td>\n        <td>0.9544</td>\n        <td>0.9838</td>\n        <td>0.9096</td>\n        <td>0.9667</td>\n    </tr>\n</table>\n\n#### Out of these techniques, K-Fold CV technique gives better overall accuracy and class wise accuracy.","metadata":{}},{"cell_type":"code","source":"path = '/kaggle/input/brain-tumor-mri-classification-dataset/Brain_Tumor_MRI_Image_Dataset/Training'","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:13:56.103740Z","iopub.execute_input":"2024-05-23T18:13:56.104790Z","iopub.status.idle":"2024-05-23T18:13:56.108857Z","shell.execute_reply.started":"2024-05-23T18:13:56.104756Z","shell.execute_reply":"2024-05-23T18:13:56.107936Z"},"trusted":true},"execution_count":3,"outputs":[]},{"cell_type":"code","source":"def get_dls_for_fold(train_indices, valid_indices):\n    split_idx = IndexSplitter(valid_indices)\n    dls = dblock.dataloaders(path, bs=64, splitter=split_idx)\n    return dls","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:13:58.523142Z","iopub.execute_input":"2024-05-23T18:13:58.523507Z","iopub.status.idle":"2024-05-23T18:13:58.528852Z","shell.execute_reply.started":"2024-05-23T18:13:58.523477Z","shell.execute_reply":"2024-05-23T18:13:58.527658Z"},"trusted":true},"execution_count":4,"outputs":[]},{"cell_type":"code","source":"def train_and_evaluate(dls):\n    learn = vision_learner(dls, resnet18, metrics=accuracy)  # Define your CNN architecture\n    learn.fine_tune(epochs=3, base_lr=1e-2, freeze_epochs=1)  # Fine-tune the model\n    \n    # Get predictions on the validation set\n    preds, _ = learn.get_preds(ds_idx=1)\n    \n    # Convert predictions to class indices\n    preds_classes = preds.argmax(dim=1)\n    \n    # Extract actual labels from the validation set\n    y_test = [parent_label(o) for o in dls.valid.items]\n    \n    # Convert predictions to class labels\n    pred_labels = [learn.dls.vocab[i] for i in preds_classes]\n    \n    # Calculate accuracy\n    acc = accuracy_score(y_test, pred_labels)\n    print(f'Model accuracy on validation set: {acc}')\n    return acc","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:14:00.289773Z","iopub.execute_input":"2024-05-23T18:14:00.290701Z","iopub.status.idle":"2024-05-23T18:14:00.299281Z","shell.execute_reply.started":"2024-05-23T18:14:00.290645Z","shell.execute_reply":"2024-05-23T18:14:00.298031Z"},"trusted":true},"execution_count":5,"outputs":[]},{"cell_type":"markdown","source":"# Hold-Out Cross Validation\n\nHoldout cross-validation (CV) is a simple technique used to estimate the performance of a machine learning model. In this technique, the dataset is split into two subsets: a training set and a validation set (also known as a test set or holdout set). The dataset is randomly partitioned into two subsets: the training set and the validation set. Typically, the training set contains a larger portion of the data, such as 70-80%, while the validation set contains the remaining portion, such as 20-30%. Holdout CV can be performed multiple times by randomly splitting the dataset into training and validation sets each time. The average performance across multiple iterations can provide a more robust estimate of the model's performance.\n\n#### Advantages:\n- Simple and easy to implement.\n- Fast training time, as it only trains the model on a subset of the data.\n\n#### Disadvantages:\n- High variance in estimated performance due to random selection of train/validation split.\n- Sensitive to how the data is divided.","metadata":{}},{"cell_type":"code","source":"from fastai.vision.all import *\n\n# Load Brain Tumor MRI Classification dataset (only training)\npath = Path('/kaggle/input/brain-tumor-mri-classification-dataset/Brain_Tumor_MRI_Image_Dataset/Training')\n\n# Create a DataBlock\ndblock = DataBlock(blocks=(ImageBlock, CategoryBlock),\n                   get_items=get_image_files,\n                   get_y=parent_label,\n                   splitter=RandomSplitter(valid_pct=0.2, seed=42),\n                   item_tfms=Resize(256),\n                   batch_tfms=aug_transforms())\n\n# Create DataLoaders\ndls = dblock.dataloaders(path, bs=64)\n\n# Train and fine tune the model\nlearn = cnn_learner(dls, resnet18, metrics=accuracy)\nlearn.fine_tune(epochs=3, base_lr=1e-3, freeze_epochs=1)\n\n# Evaluate the model\ninterp = ClassificationInterpretation.from_learner(learn)\ninterp.plot_confusion_matrix()","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:14:03.632580Z","iopub.execute_input":"2024-05-23T18:14:03.633348Z","iopub.status.idle":"2024-05-23T18:15:45.203130Z","shell.execute_reply.started":"2024-05-23T18:14:03.633317Z","shell.execute_reply":"2024-05-23T18:15:45.201895Z"},"trusted":true},"execution_count":6,"outputs":[{"name":"stderr","text":"Downloading: \"https://download.pytorch.org/models/resnet18-f37072fd.pth\" to /root/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth\n100%|██████████| 44.7M/44.7M [00:00<00:00, 138MB/s] \n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.983122</td>\n      <td>0.355160</td>\n      <td>0.863398</td>\n      <td>00:18</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.422748</td>\n      <td>0.202174</td>\n      <td>0.929947</td>\n      <td>00:20</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.285309</td>\n      <td>0.160441</td>\n      <td>0.939580</td>\n      <td>00:20</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.207024</td>\n      <td>0.154596</td>\n      <td>0.940455</td>\n      <td>00:20</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 640x480 with 1 Axes>","image/png":"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"},"metadata":{}}]},{"cell_type":"code","source":"# Get predictions and true labels\npreds, targets = learn.get_preds(dl=dls.valid)\n\n# Get predicted labels\npred_labels = preds.argmax(dim=1)\n\n# Calculate accuracy for each class\nclass_correct = {}\nclass_total = {}\nfor pred, target in zip(pred_labels, targets):\n    if target.item() not in class_correct:\n        class_correct[target.item()] = 0\n        class_total[target.item()] = 0\n    class_correct[target.item()] += (pred == target).float().sum().item()\n    class_total[target.item()] += 1\n\n# Calculate accuracy for each class\nclass_accuracies = {}\nfor cls in class_correct:\n    class_accuracies[learn.dls.vocab[cls]] = class_correct[cls] / class_total[cls]\n\n# Calculate average accuracy\naverage_accuracy = sum(class_accuracies.values()) / len(class_accuracies)\n\nprint(f'Per-class accuracies: {class_accuracies}')\nprint(f'Average accuracy across all classes: {average_accuracy:.4f}')","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:15:45.205886Z","iopub.execute_input":"2024-05-23T18:15:45.206699Z","iopub.status.idle":"2024-05-23T18:15:47.966214Z","shell.execute_reply.started":"2024-05-23T18:15:45.206641Z","shell.execute_reply":"2024-05-23T18:15:47.964695Z"},"trusted":true},"execution_count":7,"outputs":[{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Per-class accuracies: {'pituitary': 0.9857651245551602, 'notumor': 0.9758308157099698, 'meningioma': 0.8208955223880597, 'glioma': 0.9694656488549618}\nAverage accuracy across all classes: 0.9380\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# K-Fold Cross-Validation (K-Fold CV)\n\nDivides the dataset into k equal-sized folds, trains the model k times, each time using k-1 folds as training data and one fold as validation data.\n\n#### Advantages:\n- Provides more accurate and reliable estimates of model performance compared to hold-out.\n- Utilizes the entire dataset for training and validation.\n#### Disadvantages:\n- Increased computational cost due to multiple model trainings.\n- Can be slower for large datasets and complex models.","metadata":{}},{"cell_type":"code","source":"from fastai.vision.all import *\nfrom sklearn.metrics import accuracy_score, confusion_matrix\nfrom sklearn.model_selection import KFold\n\n# Load Brain Tumor MRI Classification dataset (only training)\npath = Path('/kaggle/input/brain-tumor-mri-classification-dataset/Brain_Tumor_MRI_Image_Dataset/Training')\n\n# Fastai DataBlock\ndblock = DataBlock(blocks=(ImageBlock, CategoryBlock),\n                   get_items=get_image_files,\n                   get_y=parent_label,\n                   item_tfms=Resize(256))\n\n# number of folds for K-Fold CV\nk = 5  # This k value can be changed as per out requirements/ hyperparameter tuning\n\n# Define K-Fold splitter\nkf = KFold(n_splits=k, shuffle=True, random_state=42)\n\n# Function to generate DataLoaders for each fold\ndef get_dls_for_fold(train_indices, valid_indices):\n    split_idx = IndexSplitter(valid_indices)\n    dls = dblock.dataloaders(path, bs=64, splitter=split_idx)\n    return dls\n\n# Train and evaluate the model for each fold\ndef train_and_evaluate(dls):\n    learn = vision_learner(dls, resnet18, metrics=accuracy)  # Define your CNN architecture\n    learn.fine_tune(epochs=3, base_lr=1e-3, freeze_epochs=1)  # Fine-tune the model\n    \n    # Get predictions on the validation set\n    preds, _ = learn.get_preds(ds_idx=1)\n    \n    # Convert predictions to class indices\n    preds_classes = preds.argmax(dim=1)\n    \n    # Extract actual labels from the validation set\n    y_test = [parent_label(o) for o in dls.valid.items]\n    \n    # Convert predictions to class labels\n    pred_labels = [learn.dls.vocab[i] for i in preds_classes]\n    \n    # Calculate accuracy\n    acc = accuracy_score(y_test, pred_labels)\n    print(f'Model accuracy on validation set: {acc}')\n    \n    # Calculate confusion matrix\n    cm = confusion_matrix(y_test, pred_labels, labels=learn.dls.vocab)\n    print('Confusion Matrix:')\n    print(cm)\n    \n    # Calculate per-class accuracies\n    per_class_accuracies = {cls: cm[i, i] / cm[i, :].sum() for i, cls in enumerate(learn.dls.vocab)}\n    print('Per-class accuracies:', per_class_accuracies)\n    \n    return acc, per_class_accuracies\n\n# Performing K-Fold Cross-Validation\naccuracies = []\nper_class_accuracies = {}\nfor train_indices, valid_indices in kf.split(np.arange(len(dblock.get_items(path)))):\n    dls = get_dls_for_fold(train_indices, valid_indices)\n    acc, per_class_acc = train_and_evaluate(dls)\n    accuracies.append(acc)\n    for cls, val in per_class_acc.items():\n        per_class_accuracies[cls] = per_class_accuracies.get(cls, []) + [val]\n\nprint(f'Mean accuracy over {k} folds: {np.mean(accuracies)}')\nprint('Average accuracy across all classes:', {cls: np.mean(accs) for cls, accs in per_class_accuracies.items()})","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:15:47.968283Z","iopub.execute_input":"2024-05-23T18:15:47.968704Z","iopub.status.idle":"2024-05-23T18:22:18.019206Z","shell.execute_reply.started":"2024-05-23T18:15:47.968664Z","shell.execute_reply":"2024-05-23T18:22:18.017915Z"},"trusted":true},"execution_count":8,"outputs":[{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.874660</td>\n      <td>0.335748</td>\n      <td>0.878284</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.340696</td>\n      <td>0.185882</td>\n      <td>0.929947</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.151337</td>\n      <td>0.148914</td>\n      <td>0.950088</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.071042</td>\n      <td>0.148642</td>\n      <td>0.949212</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9492119089316988\nConfusion Matrix:\n[[241  23   0   1]\n [  6 233   5  13]\n [  0   4 310   1]\n [  0   5   0 300]]\nPer-class accuracies: {'glioma': 0.909433962264151, 'meningioma': 0.9066147859922179, 'notumor': 0.9841269841269841, 'pituitary': 0.9836065573770492}\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.878811</td>\n      <td>0.367263</td>\n      <td>0.871278</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.347669</td>\n      <td>0.155526</td>\n      <td>0.941331</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.176812</td>\n      <td>0.124712</td>\n      <td>0.957093</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.084967</td>\n      <td>0.105955</td>\n      <td>0.963222</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9632224168126094\nConfusion Matrix:\n[[250  15   0   0]\n [ 11 248   6   5]\n [  0   1 306   1]\n [  0   2   1 296]]\nPer-class accuracies: {'glioma': 0.9433962264150944, 'meningioma': 0.9185185185185185, 'notumor': 0.9935064935064936, 'pituitary': 0.9899665551839465}\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.914320</td>\n      <td>0.320107</td>\n      <td>0.892294</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.343205</td>\n      <td>0.171296</td>\n      <td>0.939580</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.167071</td>\n      <td>0.124998</td>\n      <td>0.958844</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.073472</td>\n      <td>0.119346</td>\n      <td>0.959720</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9597197898423818\nConfusion Matrix:\n[[256   5   0   0]\n [ 14 235   3  15]\n [  1   4 328   1]\n [  0   2   1 277]]\nPer-class accuracies: {'glioma': 0.9808429118773946, 'meningioma': 0.8801498127340824, 'notumor': 0.9820359281437125, 'pituitary': 0.9892857142857143}\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.851312</td>\n      <td>0.288622</td>\n      <td>0.899300</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.334055</td>\n      <td>0.169288</td>\n      <td>0.943082</td>\n      <td>00:20</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.163251</td>\n      <td>0.127787</td>\n      <td>0.957968</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.078864</td>\n      <td>0.110160</td>\n      <td>0.961471</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9614711033274956\nConfusion Matrix:\n[[231  16   0   0]\n [  6 258   3   6]\n [  0   6 324   2]\n [  0   5   0 285]]\nPer-class accuracies: {'glioma': 0.9352226720647774, 'meningioma': 0.945054945054945, 'notumor': 0.9759036144578314, 'pituitary': 0.9827586206896551}\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.939101</td>\n      <td>0.304562</td>\n      <td>0.898424</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.306447</td>\n      <td>0.172837</td>\n      <td>0.945709</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.159208</td>\n      <td>0.124723</td>\n      <td>0.958844</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.073662</td>\n      <td>0.120710</td>\n      <td>0.968476</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.968476357267951\nConfusion Matrix:\n[[252  16   0   0]\n [  3 237   6   4]\n [  0   1 320   1]\n [  0   5   0 297]]\nPer-class accuracies: {'glioma': 0.9402985074626866, 'meningioma': 0.948, 'notumor': 0.9937888198757764, 'pituitary': 0.9834437086092715}\nMean accuracy over 5 folds: 0.9604203152364275\nAverage accuracy across all classes: {'glioma': 0.9418388560168207, 'meningioma': 0.9196676124599528, 'notumor': 0.9858723680221596, 'pituitary': 0.9858122312291273}\n","output_type":"stream"}]},{"cell_type":"code","source":"# Evaluate the model: Confusion Matrix\ninterp = ClassificationInterpretation.from_learner(learn)\ninterp.plot_confusion_matrix()","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:22:18.022634Z","iopub.execute_input":"2024-05-23T18:22:18.023096Z","iopub.status.idle":"2024-05-23T18:22:23.787996Z","shell.execute_reply.started":"2024-05-23T18:22:18.023059Z","shell.execute_reply":"2024-05-23T18:22:23.786370Z"},"trusted":true},"execution_count":9,"outputs":[{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 640x480 with 1 Axes>","image/png":"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"},"metadata":{}}]},{"cell_type":"code","source":"# Get predictions and true labels\npreds, targets = learn.get_preds(dl=dls.valid)\n\n# Get predicted labels\npred_labels = preds.argmax(dim=1)\n\n# Calculate accuracy for individual class\nclass_correct = {}\nclass_total = {}\nfor pred, target in zip(pred_labels, targets):\n    if target.item() not in class_correct:\n        class_correct[target.item()] = 0\n        class_total[target.item()] = 0\n    class_correct[target.item()] += (pred == target).float().sum().item()\n    class_total[target.item()] += 1\n\n# Calculate accuracy for each class\nclass_accuracies = {}\nfor cls in class_correct:\n    class_accuracies[learn.dls.vocab[cls]] = class_correct[cls] / class_total[cls]\n\n# Calculate average accuracy\naverage_accuracy = sum(class_accuracies.values()) / len(class_accuracies)\n\nprint(f'Per-class accuracies: {class_accuracies}')\nprint(f'Average accuracy across all classes: {average_accuracy:.4f}')","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:22:23.794278Z","iopub.execute_input":"2024-05-23T18:22:23.797515Z","iopub.status.idle":"2024-05-23T18:22:27.048510Z","shell.execute_reply.started":"2024-05-23T18:22:23.797473Z","shell.execute_reply":"2024-05-23T18:22:27.047315Z"},"trusted":true},"execution_count":10,"outputs":[{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Per-class accuracies: {'pituitary': 0.9834437086092715, 'notumor': 0.9968944099378882, 'glioma': 0.9776119402985075, 'meningioma': 0.892}\nAverage accuracy across all classes: 0.9625\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# Repeated K-Fold\n\nRepeats the k-fold CV process multiple times with different random splits of the data.\n\n#### Advantages:\n- Provides more robust estimates of model performance by averaging results over multiple iterations.\n\n#### Disadvantages:\n- Further increases computational cost compared to standard k-fold CV.\n\n#### Table for accuracies over 5 folds and 2 repeats\n\n<table style=\"width:90%; font-size: 16px; border-collapse: collapse; border: 1px solid black;\">\n    <tr>\n        <th rowspan=\"2\" style=\"border: 1px solid black;\">Repeats</th>\n        <th colspan=\"5\" style=\"border: 1px solid black;\">Average Accuracy across all folds</th>\n    </tr>\n    <tr>\n        <th style=\"border: 1px solid black;\">k = 1</th>\n        <th style=\"border: 1px solid black;\">k = 2</th>\n        <th style=\"border: 1px solid black;\">k = 3</th>\n        <th style=\"border: 1px solid black;\">k = 4</th>\n        <th style=\"border: 1px solid black;\">k = 5</th>\n    </tr>\n    <tr style='text-align: center;'>\n        <th style=\"border: 1px solid black;\">Repeat 1</th>\n        <td style=\"border: 1px solid black;\">0.9518</td>\n        <td style=\"border: 1px solid black;\">0.9606</td>\n        <td style=\"border: 1px solid black;\">0.9483</td>\n        <td style=\"border: 1px solid black;\">0.9527</td>\n        <td style=\"border: 1px solid black;\">0.9562</td>\n    </tr>\n    <tr style='text-align: center;'>\n        <th style=\"border: 1px solid black;\">Repeat 2</th>\n        <td style=\"border: 1px solid black;\">0.9544</td>\n        <td style=\"border: 1px solid black;\">0.9711</td>\n        <td style=\"border: 1px solid black;\">0.9685</td>\n        <td style=\"border: 1px solid black;\">0.9685</td>\n        <td style=\"border: 1px solid black;\">0.9553</td>\n    </tr>\n</table>\n","metadata":{}},{"cell_type":"code","source":"from sklearn.metrics import accuracy_score, classification_report\nfrom fastai.vision.all import *\n\n# Function to train and evaluate the model for each k fold\ndef train_and_evaluate(dls):\n    learn = vision_learner(dls, resnet18, metrics=accuracy)  # CNN architecture\n    learn.fine_tune(epochs=3, base_lr=1e-3, freeze_epochs=1)  # Fine-tune\n    \n    # Get predictions on the validation set\n    preds, _ = learn.get_preds(ds_idx=1)\n    \n    preds_classes = preds.argmax(dim=1)\n    \n    y_true = [parent_label(o) for o in dls.valid.items]\n    \n    y_pred = [learn.dls.vocab[i] for i in preds_classes] # class labels\n    \n    # Calculate accuracy\n    acc = accuracy_score(y_true, y_pred)\n    print(f'Model accuracy on validation set: {acc}')\n    \n    # Calculate per-class accuracy\n    report = classification_report(y_true, y_pred, output_dict=True)\n    per_class_acc = {k: v['precision'] for k, v in report.items() if k in learn.dls.vocab}\n    \n    return acc, per_class_acc, y_true, y_pred","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:28:38.156087Z","iopub.execute_input":"2024-05-23T18:28:38.156491Z","iopub.status.idle":"2024-05-23T18:28:38.167882Z","shell.execute_reply.started":"2024-05-23T18:28:38.156454Z","shell.execute_reply":"2024-05-23T18:28:38.166789Z"},"trusted":true},"execution_count":15,"outputs":[]},{"cell_type":"code","source":"from sklearn.model_selection import RepeatedKFold\n\n# number of folds and repeats for Repeated K-Fold CV\nk = 5  # number of folds: can be changes as per our requirements/hyperparameter tuning\nn_repeats = 2  # number of repeats\nrandom_state = 42\n\n# Repeated K-Fold splitter\nrkf = RepeatedKFold(n_splits=k, n_repeats=n_repeats, random_state=random_state)\n\n# Repeated K-Fold Cross-Validation\naccuracies = []\nper_class_accuracies = {}\nconfusion_matrices = []  # Store confusion matrices for each fold\nfor train_indices, valid_indices in rkf.split(np.arange(len(get_image_files(path)))):\n    dls = get_dls_for_fold(train_indices, valid_indices)\n    acc, per_class_acc, y_true, y_pred = train_and_evaluate(dls)\n    accuracies.append(acc)\n    for cls, val in per_class_acc.items():\n        per_class_accuracies[cls] = per_class_accuracies.get(cls, []) + [val]\n    \n    # confusion matrix for each fold and repeat\n    cm = confusion_matrix(y_true, y_pred, labels=dls.vocab)\n    confusion_matrices.append(cm)\n\nprint(f'Mean accuracy over {k} folds and {n_repeats} repeats: {np.mean(accuracies)}')\nprint('Average accuracy across all classes:', {cls: np.mean(accs) for cls, accs in per_class_accuracies.items()})","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:28:41.750350Z","iopub.execute_input":"2024-05-23T18:28:41.751290Z","iopub.status.idle":"2024-05-23T18:41:40.569678Z","shell.execute_reply.started":"2024-05-23T18:28:41.751249Z","shell.execute_reply":"2024-05-23T18:41:40.568276Z"},"trusted":true},"execution_count":16,"outputs":[{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.927386</td>\n      <td>0.391773</td>\n      <td>0.883538</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.331890</td>\n      <td>0.219358</td>\n      <td>0.934326</td>\n      <td>00:20</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.149091</td>\n      <td>0.180868</td>\n      <td>0.946585</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.076283</td>\n      <td>0.166910</td>\n      <td>0.951839</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9518388791593695\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.831081</td>\n      <td>0.332055</td>\n      <td>0.883538</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.335378</td>\n      <td>0.169845</td>\n      <td>0.945709</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.155675</td>\n      <td>0.130412</td>\n      <td>0.952715</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.075373</td>\n      <td>0.123876</td>\n      <td>0.960595</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9605954465849387\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.887765</td>\n      <td>0.327670</td>\n      <td>0.886165</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.325123</td>\n      <td>0.191464</td>\n      <td>0.933450</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.164001</td>\n      <td>0.166841</td>\n      <td>0.947461</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.075410</td>\n      <td>0.150259</td>\n      <td>0.948336</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9483362521891419\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.836924</td>\n      <td>0.353537</td>\n      <td>0.876532</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.303429</td>\n      <td>0.209265</td>\n      <td>0.927320</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.148602</td>\n      <td>0.161820</td>\n      <td>0.944834</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.070310</td>\n      <td>0.152012</td>\n      <td>0.952715</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9527145359019265\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.892513</td>\n      <td>0.339135</td>\n      <td>0.874781</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.328436</td>\n      <td>0.200757</td>\n      <td>0.928196</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.158093</td>\n      <td>0.121819</td>\n      <td>0.953590</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.073712</td>\n      <td>0.112010</td>\n      <td>0.956217</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9562171628721541\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.865760</td>\n      <td>0.307427</td>\n      <td>0.890543</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.337747</td>\n      <td>0.173772</td>\n      <td>0.941331</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.163012</td>\n      <td>0.136121</td>\n      <td>0.957093</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.070422</td>\n      <td>0.132450</td>\n      <td>0.954466</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9544658493870403\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.874036</td>\n      <td>0.273401</td>\n      <td>0.907180</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.333165</td>\n      <td>0.132472</td>\n      <td>0.951839</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.162187</td>\n      <td>0.094839</td>\n      <td>0.966725</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.076412</td>\n      <td>0.086275</td>\n      <td>0.971103</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9711033274956217\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.819699</td>\n      <td>0.296197</td>\n      <td>0.895797</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.333520</td>\n      <td>0.151991</td>\n      <td>0.942207</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.154630</td>\n      <td>0.113349</td>\n      <td>0.964974</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.073308</td>\n      <td>0.105840</td>\n      <td>0.965849</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9658493870402802\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.816408</td>\n      <td>0.299998</td>\n      <td>0.894921</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.328952</td>\n      <td>0.175537</td>\n      <td>0.947461</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.157011</td>\n      <td>0.119962</td>\n      <td>0.968476</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.075143</td>\n      <td>0.122440</td>\n      <td>0.968476</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.968476357267951\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.869713</td>\n      <td>0.313012</td>\n      <td>0.885289</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.327476</td>\n      <td>0.167768</td>\n      <td>0.937828</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.159842</td>\n      <td>0.132373</td>\n      <td>0.953590</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.077982</td>\n      <td>0.133435</td>\n      <td>0.955342</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9553415061295972\nMean accuracy over 5 folds and 2 repeats: 0.9584938704028021\nAverage accuracy across all classes: {'glioma': 0.9703241198317392, 'meningioma': 0.9255884020066073, 'notumor': 0.9838723338861992, 'pituitary': 0.95045790443823}\n","output_type":"stream"}]},{"cell_type":"code","source":"# number of folds (k) and repeats (n_repeats)\nk_values = [i for i in range(1, 6)]\nn_repeats_values = [1, 2]\n\n# Create subplots for different combinations of k and n_repeats\nfig, axes = plt.subplots(len(k_values), len(n_repeats_values), figsize=(15, 20))\n\nfor i, k in enumerate(k_values):\n    for j, n_repeats in enumerate(n_repeats_values):\n        \n        # Get the confusion matrix for the corresponding fold and repeat\n        cm = confusion_matrices[i * len(n_repeats_values) + j]  # Adjust index for confusion_matrices\n        \n        # Plot the confusion matrix\n        ax = axes[i, j]\n        im = ax.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)\n        ax.set_title(f'k={k}, n_repeats={n_repeats}')\n        plt.colorbar(im, ax=ax)\n        \n        for m in range(len(dls.vocab)):\n            for n in range(len(dls.vocab)):\n                ax.text(n, m, format(cm[m, n], '.2f'), horizontalalignment='center', verticalalignment='center', color='black')\n\n        ax.set_xticks(np.arange(len(dls.vocab)))\n        ax.set_yticks(np.arange(len(dls.vocab)))\n        ax.set_xticklabels(dls.vocab)\n        ax.set_yticklabels(dls.vocab)\n        ax.set_xlabel('Predicted labels')\n        ax.set_ylabel('True labels')\n\nplt.tight_layout()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:48:18.140370Z","iopub.execute_input":"2024-05-23T18:48:18.141010Z","iopub.status.idle":"2024-05-23T18:48:21.006287Z","shell.execute_reply.started":"2024-05-23T18:48:18.140967Z","shell.execute_reply":"2024-05-23T18:48:21.005133Z"},"trusted":true},"execution_count":27,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1500x2000 with 20 Axes>","image/png":"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"},"metadata":{}}]},{"cell_type":"markdown","source":"# Leave One Out\n\nA special case of k-fold CV where k is equal to the number of samples in the dataset. Each sample is used as a validation set once.\n\n#### Advantages:\n- Provides least biased estimate of model performance.\n\n#### Disadvantages:\n- Extremely computationally expensive, especially for large datasets.\n- High variance in performance estimate due to the single-sample validation sets.\n\n\nFor LOO CV, each data point is held out once as the validation set, and the model is trained on the remaining data points. So, if we have n data points, we will have to train the model  n times. Since LOO CV trains the model n times, the training time for each iteration depends on the size of the dataset, model architecture, and hardware resources. It can be significantly higher compared to other CV methods, especially for large datasets. In LOO CV, each data point is used once as the validation set, and the model is trained on the remaining n−1 data points. This process is repeated for each data point in dataset, resulting in n iterations, where n is the number of data points in your dataset. Therefore, for n images, we will indeed need to train the model n times in LOO CV and we have more than 5000 images for training. For this method is not feasible and will take a lot of time to execute.","metadata":{}},{"cell_type":"markdown","source":"# Stratified K-Fold\n\nLike k-fold CV, but ensures that each fold preserves the percentage of samples for each class.\n\n#### Advantages:\n- Particularly useful for imbalanced datasets, providing more representative validation sets.\n\n#### Disadvantages:\n- Slightly higher computational cost compared to standard k-fold CV due to the extra stratification step.","metadata":{}},{"cell_type":"code","source":"from sklearn.metrics import accuracy_score, classification_report\nfrom fastai.vision.all import *\n\n# train and evaluate the model for each fold\ndef train_and_evaluate(dls):\n    learn = vision_learner(dls, resnet18, metrics=accuracy)  # CNN architecture\n    learn.fine_tune(epochs=3, base_lr=1e-3, freeze_epochs=1)  # Fine-tune \n    \n    preds, _ = learn.get_preds(ds_idx=1)\n    \n    preds_classes = preds.argmax(dim=1)\n    \n    y_true = [parent_label(o) for o in dls.valid.items]\n    \n    y_pred = [learn.dls.vocab[i] for i in preds_classes]\n    \n    acc = accuracy_score(y_true, y_pred)\n    print(f'Model accuracy on validation set: {acc}')\n    \n    report = classification_report(y_true, y_pred, output_dict=True)\n    per_class_acc = {k: v['precision'] for k, v in report.items() if k in learn.dls.vocab}\n    \n    return acc, per_class_acc, y_true, y_pred","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:52:06.652239Z","iopub.execute_input":"2024-05-23T18:52:06.652685Z","iopub.status.idle":"2024-05-23T18:52:06.664037Z","shell.execute_reply.started":"2024-05-23T18:52:06.652624Z","shell.execute_reply":"2024-05-23T18:52:06.662398Z"},"trusted":true},"execution_count":30,"outputs":[]},{"cell_type":"code","source":"\nfrom sklearn.model_selection import StratifiedKFold\n\n# Number of folds for Stratified K-Fold CV\nk = 5  # number of folds\n\n# Stratified K-Fold splitter\nskf = StratifiedKFold(n_splits=k, shuffle=True, random_state=42)\n\n# Stratified K-Fold Cross-Validation\naccuracies = []\nper_class_accuracies = {}\nconfusion_matrices = []  # Store confusion matrices for each fold\nfor train_indices, valid_indices in skf.split(np.arange(len(get_image_files(path))), [parent_label(o) for o in get_image_files(path)]):\n    dls = get_dls_for_fold(train_indices, valid_indices)\n    acc, per_class_acc, y_true, y_pred = train_and_evaluate(dls)\n    accuracies.append(acc)\n    for cls, val in per_class_acc.items():\n        per_class_accuracies[cls] = per_class_accuracies.get(cls, []) + [val]\n    \n    # confusion matrix\n    cm = confusion_matrix(y_true, y_pred, labels=dls.vocab)\n    confusion_matrices.append(cm)\n\nprint(f'Mean accuracy over {k} folds: {np.mean(accuracies)}')\nprint('Average accuracy across all classes:', {cls: np.mean(accs) for cls, accs in per_class_accuracies.items()})","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:52:17.791207Z","iopub.execute_input":"2024-05-23T18:52:17.791599Z","iopub.status.idle":"2024-05-23T18:58:49.749207Z","shell.execute_reply.started":"2024-05-23T18:52:17.791567Z","shell.execute_reply":"2024-05-23T18:58:49.747228Z"},"trusted":true},"execution_count":31,"outputs":[{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.903851</td>\n      <td>0.316393</td>\n      <td>0.891419</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.338887</td>\n      <td>0.184448</td>\n      <td>0.941331</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.156008</td>\n      <td>0.133695</td>\n      <td>0.959720</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.071222</td>\n      <td>0.133809</td>\n      <td>0.960595</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9605954465849387\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.891530</td>\n      <td>0.367875</td>\n      <td>0.885289</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.306223</td>\n      <td>0.225318</td>\n      <td>0.929072</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.145077</td>\n      <td>0.187035</td>\n      <td>0.941331</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.068084</td>\n      <td>0.173529</td>\n      <td>0.944834</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9448336252189142\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.914143</td>\n      <td>0.383797</td>\n      <td>0.873905</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.330520</td>\n      <td>0.221491</td>\n      <td>0.928196</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.150833</td>\n      <td>0.171103</td>\n      <td>0.953590</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.076399</td>\n      <td>0.164667</td>\n      <td>0.953590</td>\n      <td>00:20</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9535901926444834\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.862717</td>\n      <td>0.305703</td>\n      <td>0.895797</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.316800</td>\n      <td>0.188480</td>\n      <td>0.936953</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.158771</td>\n      <td>0.123750</td>\n      <td>0.961471</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.066293</td>\n      <td>0.121189</td>\n      <td>0.960595</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9605954465849387\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.872977</td>\n      <td>0.379292</td>\n      <td>0.881786</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.328655</td>\n      <td>0.220226</td>\n      <td>0.930823</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.167332</td>\n      <td>0.184127</td>\n      <td>0.954466</td>\n      <td>00:19</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.079687</td>\n      <td>0.167778</td>\n      <td>0.954466</td>\n      <td>00:19</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stdout","text":"Model accuracy on validation set: 0.9544658493870403\nMean accuracy over 5 folds: 0.954816112084063\nAverage accuracy across all classes: {'glioma': 0.966718847934783, 'meningioma': 0.9096564314102453, 'notumor': 0.9838569239608793, 'pituitary': 0.9544664477148981}\n","output_type":"stream"}]},{"cell_type":"code","source":"# number of folds (k)\nk_values = [i for i in range(1, 6)]\n\nfig, axes = plt.subplots(len(k_values), 1, figsize=(15, 5 * len(k_values)))\n\nfor i, k in enumerate(k_values):\n    cm = confusion_matrices[i]\n    \n    ax = axes[i]\n    im = ax.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues) # confusion matrix display\n    ax.set_title(f'k={k}')\n    plt.colorbar(im, ax=ax)\n    \n    for m in range(len(dls.vocab)):\n        for n in range(len(dls.vocab)):\n            ax.text(n, m, format(cm[m, n], '.2f'), horizontalalignment='center', verticalalignment='center', color='black')\n\n    ax.set_xticks(np.arange(len(dls.vocab)))\n    ax.set_yticks(np.arange(len(dls.vocab)))\n    ax.set_xticklabels(dls.vocab)\n    ax.set_yticklabels(dls.vocab)\n    ax.set_xlabel('Predicted labels')\n    ax.set_ylabel('True labels')\n\nplt.tight_layout()\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-05-23T18:58:58.686177Z","iopub.execute_input":"2024-05-23T18:58:58.686931Z","iopub.status.idle":"2024-05-23T18:59:00.270923Z","shell.execute_reply.started":"2024-05-23T18:58:58.686882Z","shell.execute_reply":"2024-05-23T18:59:00.269810Z"},"trusted":true},"execution_count":32,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 1500x2500 with 10 Axes>","image/png":"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"},"metadata":{}}]}]}
\ No newline at end of file
diff --git a/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md b/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md
new file mode 100644
index 00000000..41cc2603
--- /dev/null
+++ b/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md	
@@ -0,0 +1,78 @@
+# Hyper-Parameter Tuning Techniques for Brain Tumor MRI Classification
+____
+
+The dataset on which the cross validation is carried out can be found [on kaggle](https://www.kaggle.com/datasets/theiturhs/brain-tumor-mri-classification-dataset/data). Find the implementation of this [in this notebook](). Learner module, FastAI, is considered to provide a convenient way to create and fine-tune convolutional neural network (CNN) models. vision.learner is a function that helps us to construct a learner object, which has the model architecture, data, training configuration, and other elements. We can specify a pre-trained model architecture and fine-tune it on the dataset.
+
+### Hyper-Parameter Tuning Teachniques carried out are as follows:
+*Different techniques to find out suitable hyper-parameters*
+* Random Search optimization algorithm
+* Hyperparameter Optimization with Optuna's Successive Halving Pruner
+
+### Different techniques and their obtained training accuracies¶
+| Technique | Accuracy Score | Architecture | Weight Decay | Epochs | Batch Size | Drop |
+| -- | -- | -- | -- | -- | -- | -- |
+| Random Search optimization algorithm - Run 1 | 95.27% | ResNet50 | 5.527e-5 | 7 | 64 | 0.4 |
+| Random Search optimization algorithm - Run 2 | 95.53% | ResNet50 | 4e-6 | 5 | 64 | 0.2 |
+| Hyperparameter Optimization with Optuna's Successive Halving Pruner | 
+
+**Out of these techniques,Hyperparameter Optimization with Optuna's Successive Halving Pruner technique gives better overall accuracy. It almost take 40-50 minutes for each to fine-tune the model.**
+
+NOTE: Since here we need to find the best technique that we can use for hyper-parameter tuning of our dataset, and it takes almost 40-50 minutes on an average to get results from each techniques (where we have only considered training dataset not augmented data), thus RandomSplitting is implemented to divide the dataset.
+
+#### Random Search Optimizing Algorithm - Run 1
+
+For n_trails = 10, the accuracy score and best hyper-parameter are as follows:
+
+| Trail No. | Best Score | Architecture | Weight Decay | Epochs | Batch Size | Drop |
+|-----------|------------|--------------|--------------|--------|------------|------|
+| 0         | 0.9011     | ResNet34     | 0.00024      | 8      | 64         | 0.4  |
+| 1         | 0.9413     | ResNet18     | 0.0090       | 15     | 64         | 0.2  |
+| 2         | 0.9343     | ResNet18     | 0.0065       | 5      | 32         | 0.4  |
+| 3         | 0.9080     | ResNet34     | 0.00062      | 5      | 32         | 0.4  |
+| 4         | 0.9019     | ResNet34     | 0.00092      | 6      | 64         | 0.2  |
+| **5**         | **0.9527**     | **ResNet50**     | **0.00005**      | **7**      | **64**         | **0.4**  |
+| 6         | 0.9220     | ResNet34     | 0.00895      | 15     | 64         | 0.2  |
+| 7         | 0.9404     | ResNet18     | 0.0035       | 11     | 64         | 0.4  |
+| 8         | 0.9212     | ResNet18     | 0.0002       | 5      | 64         | 0.4  |
+| 9         | 0.9203     | ResNet34     | 0.0007       | 13     | 32         | 0.4  |
+
+#### Random Search Optimizing Algorithm - Run 2
+
+For n_trails = 10, the accuracy score and best hyper-parameter are as follows:
+
+| Trial | Best Score | Architecture | Weight Decay | Epochs | Batch Size | Drop |
+|-------|------------|--------------|--------------|--------|------------|------|
+| 0     | 0.9177     | ResNet34     | 0.000248     | 6      | 32         | 0.2  |
+| 1     | 0.9492     | ResNet50     | 0.002137     | 7      | 64         | 0.4  |
+| 2     | 0.9518     | ResNet50     | 0.000004     | 8      | 64         | 0.2  |
+| 3     | 0.9046     | ResNet34     | 0.000269     | 6      | 64         | 0.2  |
+| 4     | 0.9378     | ResNet50     | 0.000058     | 6      | 32         | 0.2  |
+| 5     | 0.9352     | ResNet18     | 0.000154     | 7      | 32         | 0.2  |
+| 6     | 0.9063     | ResNet34     | 0.000006     | 5      | 64         | 0.4  |
+| **7**     | **0.9553**     | **ResNet50**     | **0.000004**     | **13**     | **64**         | **0.2**  |
+| 8     | 0.9238     | ResNet34     | 0.000824     | 13     | 64         | 0.4  |
+| 9     | 0.9361     | ResNet18     | 0.000018     | 8      | 32         | 0.2  |
+
+#### Hyperparameter Optimization with Optuna's Successive Halving Pruner
+
+For n_trails = 10, accuracy scores and hyper-paramters are:
+
+| Trial | Best Score | Architecture | Weight Decay | Epochs | Batch Size | Drop               |
+|-------|------------|--------------|--------------|--------|------------|--------------------|
+| 0     | 0.9623     | resnet50     | 2.057e-06    | 8      | 32         | 0.2888             |
+| 1     | 0.9518     | resnet50     | 0.001421     | 7      | 32         | 0.2707             |
+| 2     | 0.9807     | resnet34     | 3.744e-06    | 9      | 32         | 0.3918             |
+| 3     | 0.9641     | resnet18     | 2.667e-05    | 7      | 64         | 0.3196             |
+| 4     | 0.9711     | resnet34     | 0.005883     | 8      | 64         | 0.2000             |
+| 5     | 0.9650     | resnet18     | 5.694e-06    | 6      | 64         | 0.2266             |
+| 6     | 0.9737     | resnet18     | 1.813e-05    | 6      | 64         | 0.3732             |
+| **7**     | **0.9851**     | **resnet34**     | **0.004016**     | **13**     | **32**         | **0.2680**             |
+| 8     | 0.9667     | resnet50     | 0.008095     | 15     | 32         | 0.3013             |
+| 9     | 0.9632     | resnet50     | 0.0006995    | 6      | 32         | 0.3595             |
+
+##### Summarizing
+
+1. Random search optimization gives almost 95% accuracy when we ran it two times each with 10 iterations.
+2. Hyperparameter Optimization with Optuna's Successive Halving Pruner, we are getting 98.51% which is a remarkable accuracy.
+
+**So we will be using k-fold validation technique followed by Hyperparameter Optimization with Optuna's Successive Halving Pruner for getting the appropriate hyper-parameters.**
\ No newline at end of file
diff --git a/Brain Tumor MRI Classification/HyperParameter_Tuning.ipynb b/Brain Tumor MRI Classification/HyperParameter_Tuning.ipynb
new file mode 100644
index 00000000..2b62487d
--- /dev/null
+++ b/Brain Tumor MRI Classification/HyperParameter_Tuning.ipynb	
@@ -0,0 +1 @@
+{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"nvidiaTeslaT4","dataSources":[{"sourceId":8392068,"sourceType":"datasetVersion","datasetId":4992016}],"dockerImageVersionId":30699,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Hyper-Parameter Tuning\n\nThis notebook contains hyperparameter tuning for the model. Learner module, FastAI, provides a convenient way to create and fine-tune convolutional neural network (CNN) models. vision.learner is a function that helps us to construct a learner object, which has the model architecture, data, training configuration, and other elements. We can specify a pre-trained model architecture and fine-tune it on the dataset. vision.learner supports a wide range of CNN architectures.\n\nThis notebook includes these following implementations:\n\n1. Random Search optimization algorithm - Run 1\n2. Random Search optimization algorithm - Run 2\n3. Hyperparameter Optimization with Optuna's Successive Halving Pruner","metadata":{}},{"cell_type":"code","source":"import warnings\nwarnings.filterwarnings(\"ignore\")","metadata":{"execution":{"iopub.status.busy":"2024-05-24T21:49:21.580727Z","iopub.execute_input":"2024-05-24T21:49:21.581162Z","iopub.status.idle":"2024-05-24T21:49:21.593364Z","shell.execute_reply.started":"2024-05-24T21:49:21.581118Z","shell.execute_reply":"2024-05-24T21:49:21.592427Z"},"trusted":true},"execution_count":1,"outputs":[]},{"cell_type":"code","source":"!pip install optuna-integration","metadata":{"scrolled":true,"execution":{"iopub.status.busy":"2024-05-24T21:49:21.595328Z","iopub.execute_input":"2024-05-24T21:49:21.595586Z","iopub.status.idle":"2024-05-24T21:49:35.372636Z","shell.execute_reply.started":"2024-05-24T21:49:21.595558Z","shell.execute_reply":"2024-05-24T21:49:35.371477Z"},"trusted":true},"execution_count":2,"outputs":[{"name":"stdout","text":"Collecting optuna-integration\n  Downloading optuna_integration-3.6.0-py3-none-any.whl.metadata (10 kB)\nRequirement already satisfied: optuna in /opt/conda/lib/python3.10/site-packages (from optuna-integration) (3.6.1)\nRequirement already satisfied: alembic>=1.5.0 in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (1.13.1)\nRequirement already satisfied: colorlog in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (6.8.2)\nRequirement already satisfied: numpy in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (1.26.4)\nRequirement already satisfied: packaging>=20.0 in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (21.3)\nRequirement already satisfied: sqlalchemy>=1.3.0 in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (2.0.25)\nRequirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (4.66.1)\nRequirement already satisfied: PyYAML in /opt/conda/lib/python3.10/site-packages (from optuna->optuna-integration) (6.0.1)\nRequirement already satisfied: Mako in /opt/conda/lib/python3.10/site-packages (from alembic>=1.5.0->optuna->optuna-integration) (1.3.3)\nRequirement already satisfied: typing-extensions>=4 in /opt/conda/lib/python3.10/site-packages (from alembic>=1.5.0->optuna->optuna-integration) (4.9.0)\nRequirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /opt/conda/lib/python3.10/site-packages (from packaging>=20.0->optuna->optuna-integration) (3.1.1)\nRequirement already satisfied: greenlet!=0.4.17 in /opt/conda/lib/python3.10/site-packages (from sqlalchemy>=1.3.0->optuna->optuna-integration) (3.0.3)\nRequirement already satisfied: MarkupSafe>=0.9.2 in /opt/conda/lib/python3.10/site-packages (from Mako->alembic>=1.5.0->optuna->optuna-integration) (2.1.3)\nDownloading optuna_integration-3.6.0-py3-none-any.whl (93 kB)\n\u001b[2K   \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m93.4/93.4 kB\u001b[0m \u001b[31m2.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n\u001b[?25hInstalling collected packages: optuna-integration\nSuccessfully installed optuna-integration-3.6.0\n","output_type":"stream"}]},{"cell_type":"code","source":"!pip install optuna lightgbm","metadata":{"scrolled":true,"execution":{"iopub.status.busy":"2024-05-24T21:49:35.374410Z","iopub.execute_input":"2024-05-24T21:49:35.374788Z","iopub.status.idle":"2024-05-24T21:49:47.583346Z","shell.execute_reply.started":"2024-05-24T21:49:35.374753Z","shell.execute_reply":"2024-05-24T21:49:47.582208Z"},"trusted":true},"execution_count":3,"outputs":[{"name":"stdout","text":"Requirement already satisfied: optuna in /opt/conda/lib/python3.10/site-packages (3.6.1)\nRequirement already satisfied: lightgbm in /opt/conda/lib/python3.10/site-packages (4.2.0)\nRequirement already satisfied: alembic>=1.5.0 in /opt/conda/lib/python3.10/site-packages (from optuna) (1.13.1)\nRequirement already satisfied: colorlog in /opt/conda/lib/python3.10/site-packages (from optuna) (6.8.2)\nRequirement already satisfied: numpy in /opt/conda/lib/python3.10/site-packages (from optuna) (1.26.4)\nRequirement already satisfied: packaging>=20.0 in /opt/conda/lib/python3.10/site-packages (from optuna) (21.3)\nRequirement already satisfied: sqlalchemy>=1.3.0 in /opt/conda/lib/python3.10/site-packages (from optuna) (2.0.25)\nRequirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from optuna) (4.66.1)\nRequirement already satisfied: PyYAML in /opt/conda/lib/python3.10/site-packages (from optuna) (6.0.1)\nRequirement already satisfied: scipy in /opt/conda/lib/python3.10/site-packages (from lightgbm) (1.11.4)\nRequirement already satisfied: Mako in /opt/conda/lib/python3.10/site-packages (from alembic>=1.5.0->optuna) (1.3.3)\nRequirement already satisfied: typing-extensions>=4 in /opt/conda/lib/python3.10/site-packages (from alembic>=1.5.0->optuna) (4.9.0)\nRequirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /opt/conda/lib/python3.10/site-packages (from packaging>=20.0->optuna) (3.1.1)\nRequirement already satisfied: greenlet!=0.4.17 in /opt/conda/lib/python3.10/site-packages (from sqlalchemy>=1.3.0->optuna) (3.0.3)\nRequirement already satisfied: MarkupSafe>=0.9.2 in /opt/conda/lib/python3.10/site-packages (from Mako->alembic>=1.5.0->optuna) (2.1.3)\n","output_type":"stream"}]},{"cell_type":"code","source":"!pip install --upgrade optuna","metadata":{"scrolled":true,"execution":{"iopub.status.busy":"2024-05-24T21:49:47.585635Z","iopub.execute_input":"2024-05-24T21:49:47.585943Z","iopub.status.idle":"2024-05-24T21:49:59.959461Z","shell.execute_reply.started":"2024-05-24T21:49:47.585914Z","shell.execute_reply":"2024-05-24T21:49:59.958360Z"},"trusted":true},"execution_count":4,"outputs":[{"name":"stdout","text":"Requirement already satisfied: optuna in /opt/conda/lib/python3.10/site-packages (3.6.1)\nRequirement already satisfied: alembic>=1.5.0 in /opt/conda/lib/python3.10/site-packages (from optuna) (1.13.1)\nRequirement already satisfied: colorlog in /opt/conda/lib/python3.10/site-packages (from optuna) (6.8.2)\nRequirement already satisfied: numpy in /opt/conda/lib/python3.10/site-packages (from optuna) (1.26.4)\nRequirement already satisfied: packaging>=20.0 in /opt/conda/lib/python3.10/site-packages (from optuna) (21.3)\nRequirement already satisfied: sqlalchemy>=1.3.0 in /opt/conda/lib/python3.10/site-packages (from optuna) (2.0.25)\nRequirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from optuna) (4.66.1)\nRequirement already satisfied: PyYAML in /opt/conda/lib/python3.10/site-packages (from optuna) (6.0.1)\nRequirement already satisfied: Mako in /opt/conda/lib/python3.10/site-packages (from alembic>=1.5.0->optuna) (1.3.3)\nRequirement already satisfied: typing-extensions>=4 in /opt/conda/lib/python3.10/site-packages (from alembic>=1.5.0->optuna) (4.9.0)\nRequirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /opt/conda/lib/python3.10/site-packages (from packaging>=20.0->optuna) (3.1.1)\nRequirement already satisfied: greenlet!=0.4.17 in /opt/conda/lib/python3.10/site-packages (from sqlalchemy>=1.3.0->optuna) (3.0.3)\nRequirement already satisfied: MarkupSafe>=0.9.2 in /opt/conda/lib/python3.10/site-packages (from Mako->alembic>=1.5.0->optuna) (2.1.3)\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# Random Search optimization algorithm - Run 1\n\nThe optimization algorithm used below is Optuna's default algorithm, which is a Tree-structured Parzen Estimator (TPE) algorithm.\n\nFor n_trails = 10, the accuracy score and best hyper-parameter are as follows:\n\n| Trail No. | Best Score | Architecture | Weight Decay | Epochs | Batch Size | Drop |\n|-----------|------------|--------------|--------------|--------|------------|------|\n| 0         | 0.9011     | ResNet34     | 0.00024      | 8      | 64         | 0.4  |\n| 1         | 0.9413     | ResNet18     | 0.0090       | 15     | 64         | 0.2  |\n| 2         | 0.9343     | ResNet18     | 0.0065       | 5      | 32         | 0.4  |\n| 3         | 0.9080     | ResNet34     | 0.00062      | 5      | 32         | 0.4  |\n| 4         | 0.9019     | ResNet34     | 0.00092      | 6      | 64         | 0.2  |\n| 5         | 0.9527     | ResNet50     | 0.00005      | 7      | 64         | 0.4  |\n| 6         | 0.9220     | ResNet34     | 0.00895      | 15     | 64         | 0.2  |\n| 7         | 0.9404     | ResNet18     | 0.0035       | 11     | 64         | 0.4  |\n| 8         | 0.9212     | ResNet18     | 0.0002       | 5      | 64         | 0.4  |\n| 9         | 0.9203     | ResNet34     | 0.0007       | 13     | 32         | 0.4  |\n\n\n**The best aacuracy score is 0.9527 with these hyperparameters:**\n- Architecture: ResNet 50\n- Weight Decay: 5.527e-5\n- Epochs: 7\n- Batch Size: 64\n- Drop: 0.4","metadata":{}},{"cell_type":"code","source":"import optuna\nfrom fastai.vision.all import *\nfrom sklearn.model_selection import cross_val_score\n\n# objective function for hyperparameter optimization\ndef objective(trial):\n    path = Path('/kaggle/input/brain-tumor-mri-classification-dataset/Brain_Tumor_MRI_Image_Dataset/Training')\n    \n    # hyperparameters to optimize\n    arch = trial.suggest_categorical('arch', ['resnet18', 'resnet34', 'resnet50'])\n    wd = trial.suggest_loguniform('wd', 1e-6, 1e-2)\n    epochs = trial.suggest_int('epochs', 5, 15)\n    bs = trial.suggest_categorical('bs', [32, 64])\n    drop = trial.suggest_categorical('drop', [0.2, 0.4])\n    \n    # DataBlock to prepare the data\n    dblock = DataBlock(blocks=(ImageBlock, CategoryBlock),\n                       get_items=get_image_files,\n                       get_y=parent_label,\n                       splitter=RandomSplitter(valid_pct=0.2, seed=42))\n    \n    dls = dblock.dataloaders(path, bs=bs) # Create DataLoaders\n    \n    # learner with specified architecture, metrics, weight decay, and data\n    learn = vision_learner(dls, arch, metrics=accuracy, wd=wd)\n    \n    # Fine-tune the model\n    learn.fine_tune(epochs, base_lr=0.001, cbs=[MixedPrecision()])\n    \n    return float(learn.validate()[1]) # validation accuracy of the trained model\n\n# Optuna study object for hyperparameter optimization\nstudy = optuna.create_study(direction=\"maximize\")\n\n# Optimize the objective function by running multiple trials\nstudy.optimize(objective, n_trials=10)\n\ntrial = study.best_trial # best trial from the study\n\nprint(\"Accuracy: {}\".format(trial.value))\nprint(\"Best hyperparameters: {}\".format(trial.params))","metadata":{"execution":{"iopub.status.busy":"2024-05-24T21:49:59.961085Z","iopub.execute_input":"2024-05-24T21:49:59.961627Z","iopub.status.idle":"2024-05-24T22:15:43.591342Z","shell.execute_reply.started":"2024-05-24T21:49:59.961595Z","shell.execute_reply":"2024-05-24T22:15:43.590254Z"},"trusted":true},"execution_count":5,"outputs":[{"name":"stderr","text":"[I 2024-05-24 21:50:09,794] A new study created in memory with name: no-name-10ac8b0a-751b-4f7e-b013-29bab511d22f\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"model.safetensors:   0%|          | 0.00/87.3M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"4d189e1c529449cdb50b6a61cfcb4556"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>1.096213</td>\n      <td>0.387679</td>\n      <td>0.865149</td>\n      <td>00:14</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.568723</td>\n      <td>0.322058</td>\n      <td>0.892294</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.490151</td>\n      <td>0.291415</td>\n      <td>0.892294</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.417919</td>\n      <td>0.286482</td>\n      <td>0.898424</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.361479</td>\n      <td>0.278917</td>\n      <td>0.884413</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.310722</td>\n      <td>0.246992</td>\n      <td>0.907180</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.267015</td>\n      <td>0.246769</td>\n      <td>0.904553</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.226637</td>\n      <td>0.251126</td>\n      <td>0.898424</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.210725</td>\n      <td>0.248565</td>\n      <td>0.901926</td>\n      <td>00:16</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 21:52:42,748] Trial 0 finished with value: 0.9010508060455322 and parameters: {'arch': 'resnet34', 'wd': 0.00023340994424944704, 'epochs': 8, 'bs': 64, 'drop': 0.4}. Best is trial 0 with value: 0.9010508060455322.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"model.safetensors:   0%|          | 0.00/46.8M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"50e0891ed7714cf69ff55757f5fc37db"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.930997</td>\n      <td>0.300741</td>\n      <td>0.893170</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.438969</td>\n      <td>0.262434</td>\n      <td>0.903678</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.389904</td>\n      <td>0.233369</td>\n      <td>0.918564</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.346242</td>\n      <td>0.217925</td>\n      <td>0.919440</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.301022</td>\n      <td>0.217099</td>\n      <td>0.917688</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.255345</td>\n      <td>0.200015</td>\n      <td>0.920315</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.222467</td>\n      <td>0.182874</td>\n      <td>0.930823</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.190116</td>\n      <td>0.173722</td>\n      <td>0.930823</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.150652</td>\n      <td>0.166545</td>\n      <td>0.934326</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.122943</td>\n      <td>0.173149</td>\n      <td>0.934326</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.121059</td>\n      <td>0.168385</td>\n      <td>0.929947</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.116469</td>\n      <td>0.163347</td>\n      <td>0.936077</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.103282</td>\n      <td>0.159642</td>\n      <td>0.933450</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.091288</td>\n      <td>0.153514</td>\n      <td>0.937828</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>13</td>\n      <td>0.096428</td>\n      <td>0.156008</td>\n      <td>0.942207</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>14</td>\n      <td>0.091354</td>\n      <td>0.154579</td>\n      <td>0.938704</td>\n      <td>00:11</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 21:55:42,252] Trial 1 finished with value: 0.9413309693336487 and parameters: {'arch': 'resnet18', 'wd': 0.009064248812822551, 'epochs': 15, 'bs': 64, 'drop': 0.2}. Best is trial 1 with value: 0.9413309693336487.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.770679</td>\n      <td>0.276929</td>\n      <td>0.894921</td>\n      <td>00:10</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.436824</td>\n      <td>0.254958</td>\n      <td>0.914186</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.360165</td>\n      <td>0.212044</td>\n      <td>0.922067</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.310777</td>\n      <td>0.203068</td>\n      <td>0.911559</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.258737</td>\n      <td>0.172688</td>\n      <td>0.936077</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.235185</td>\n      <td>0.161799</td>\n      <td>0.934326</td>\n      <td>00:12</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 21:56:56,249] Trial 2 finished with value: 0.9343257546424866 and parameters: {'arch': 'resnet18', 'wd': 0.006544389818968023, 'epochs': 5, 'bs': 32, 'drop': 0.4}. Best is trial 1 with value: 0.9413309693336487.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.947751</td>\n      <td>0.383075</td>\n      <td>0.863398</td>\n      <td>00:13</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.545927</td>\n      <td>0.322183</td>\n      <td>0.887916</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.442133</td>\n      <td>0.286373</td>\n      <td>0.894046</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.404942</td>\n      <td>0.258899</td>\n      <td>0.902802</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.322297</td>\n      <td>0.254092</td>\n      <td>0.904553</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.296800</td>\n      <td>0.253844</td>\n      <td>0.909807</td>\n      <td>00:17</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 21:58:42,859] Trial 3 finished with value: 0.9080560207366943 and parameters: {'arch': 'resnet34', 'wd': 0.0006280450180845069, 'epochs': 5, 'bs': 32, 'drop': 0.4}. Best is trial 1 with value: 0.9413309693336487.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>1.133325</td>\n      <td>0.413726</td>\n      <td>0.849387</td>\n      <td>00:12</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.537261</td>\n      <td>0.325078</td>\n      <td>0.886165</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.463902</td>\n      <td>0.291431</td>\n      <td>0.887916</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.401973</td>\n      <td>0.273693</td>\n      <td>0.885289</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.321756</td>\n      <td>0.254494</td>\n      <td>0.901051</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.286910</td>\n      <td>0.253121</td>\n      <td>0.900175</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.270712</td>\n      <td>0.244473</td>\n      <td>0.903678</td>\n      <td>00:16</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:00:36,999] Trial 4 finished with value: 0.9019264578819275 and parameters: {'arch': 'resnet34', 'wd': 0.0009150842875866279, 'epochs': 6, 'bs': 64, 'drop': 0.2}. Best is trial 1 with value: 0.9413309693336487.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"model.safetensors:   0%|          | 0.00/102M [00:00<?, ?B/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"038e28b6fcc840db878e40b0cb0e5ce0"}},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.826989</td>\n      <td>0.357276</td>\n      <td>0.862522</td>\n      <td>00:21</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.320762</td>\n      <td>0.251823</td>\n      <td>0.927320</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.284760</td>\n      <td>0.241625</td>\n      <td>0.930823</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.228570</td>\n      <td>0.245083</td>\n      <td>0.934326</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.191758</td>\n      <td>0.222438</td>\n      <td>0.942207</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.144681</td>\n      <td>0.172780</td>\n      <td>0.950963</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.109777</td>\n      <td>0.198920</td>\n      <td>0.954466</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.090422</td>\n      <td>0.216000</td>\n      <td>0.951839</td>\n      <td>00:26</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:04:07,216] Trial 5 finished with value: 0.9527145624160767 and parameters: {'arch': 'resnet50', 'wd': 5.527401639726816e-05, 'epochs': 7, 'bs': 64, 'drop': 0.4}. Best is trial 5 with value: 0.9527145624160767.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>1.098644</td>\n      <td>0.400075</td>\n      <td>0.852890</td>\n      <td>00:13</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.549288</td>\n      <td>0.335254</td>\n      <td>0.881786</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.506567</td>\n      <td>0.308290</td>\n      <td>0.888792</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.456383</td>\n      <td>0.285498</td>\n      <td>0.891419</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.401892</td>\n      <td>0.273000</td>\n      <td>0.900175</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.343611</td>\n      <td>0.281282</td>\n      <td>0.890543</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.282301</td>\n      <td>0.236686</td>\n      <td>0.908056</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.248601</td>\n      <td>0.246622</td>\n      <td>0.910683</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.221422</td>\n      <td>0.240875</td>\n      <td>0.911559</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.193505</td>\n      <td>0.214496</td>\n      <td>0.922942</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.180286</td>\n      <td>0.205274</td>\n      <td>0.928196</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.167603</td>\n      <td>0.214673</td>\n      <td>0.928196</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.142670</td>\n      <td>0.209276</td>\n      <td>0.913310</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.135244</td>\n      <td>0.201320</td>\n      <td>0.927320</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>13</td>\n      <td>0.129844</td>\n      <td>0.203409</td>\n      <td>0.922067</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>14</td>\n      <td>0.132031</td>\n      <td>0.205030</td>\n      <td>0.922067</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:08:28,633] Trial 6 finished with value: 0.9220665693283081 and parameters: {'arch': 'resnet34', 'wd': 0.008951501361629699, 'epochs': 15, 'bs': 64, 'drop': 0.2}. Best is trial 5 with value: 0.9527145624160767.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.905936</td>\n      <td>0.306733</td>\n      <td>0.894921</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.434151</td>\n      <td>0.277744</td>\n      <td>0.901051</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.390968</td>\n      <td>0.236190</td>\n      <td>0.915937</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.350471</td>\n      <td>0.223331</td>\n      <td>0.917688</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.292414</td>\n      <td>0.195645</td>\n      <td>0.918564</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.250865</td>\n      <td>0.187537</td>\n      <td>0.926445</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.208969</td>\n      <td>0.178643</td>\n      <td>0.927320</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.176541</td>\n      <td>0.175607</td>\n      <td>0.933450</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.151058</td>\n      <td>0.172329</td>\n      <td>0.933450</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.139511</td>\n      <td>0.166510</td>\n      <td>0.936953</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.124001</td>\n      <td>0.164608</td>\n      <td>0.935201</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.123595</td>\n      <td>0.162609</td>\n      <td>0.937828</td>\n      <td>00:10</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:10:36,906] Trial 7 finished with value: 0.9404553174972534 and parameters: {'arch': 'resnet18', 'wd': 0.003537091430333875, 'epochs': 11, 'bs': 64, 'drop': 0.4}. Best is trial 5 with value: 0.9527145624160767.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.941354</td>\n      <td>0.277954</td>\n      <td>0.893170</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.441243</td>\n      <td>0.266596</td>\n      <td>0.901051</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.376036</td>\n      <td>0.217937</td>\n      <td>0.908932</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.306597</td>\n      <td>0.209907</td>\n      <td>0.914186</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.250445</td>\n      <td>0.195011</td>\n      <td>0.926445</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.224316</td>\n      <td>0.202344</td>\n      <td>0.922942</td>\n      <td>00:10</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:11:41,417] Trial 8 finished with value: 0.9211909174919128 and parameters: {'arch': 'resnet18', 'wd': 0.00020852879448736512, 'epochs': 5, 'bs': 64, 'drop': 0.4}. Best is trial 5 with value: 0.9527145624160767.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.910820</td>\n      <td>0.359393</td>\n      <td>0.868652</td>\n      <td>00:13</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.537367</td>\n      <td>0.309434</td>\n      <td>0.897548</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.480948</td>\n      <td>0.295927</td>\n      <td>0.887040</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.435401</td>\n      <td>0.249373</td>\n      <td>0.912434</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.376859</td>\n      <td>0.250839</td>\n      <td>0.903678</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.316053</td>\n      <td>0.253130</td>\n      <td>0.906305</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.278214</td>\n      <td>0.218644</td>\n      <td>0.918564</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.260835</td>\n      <td>0.218007</td>\n      <td>0.908056</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.231998</td>\n      <td>0.208978</td>\n      <td>0.921191</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.186753</td>\n      <td>0.204182</td>\n      <td>0.922067</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.184746</td>\n      <td>0.203614</td>\n      <td>0.917688</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.180596</td>\n      <td>0.208405</td>\n      <td>0.922067</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.186264</td>\n      <td>0.197494</td>\n      <td>0.924694</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.180821</td>\n      <td>0.199968</td>\n      <td>0.918564</td>\n      <td>00:17</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:15:43,586] Trial 9 finished with value: 0.9203152656555176 and parameters: {'arch': 'resnet34', 'wd': 0.0007385132863360378, 'epochs': 13, 'bs': 32, 'drop': 0.4}. Best is trial 5 with value: 0.9527145624160767.\n","output_type":"stream"},{"name":"stdout","text":"Accuracy: 0.9527145624160767\nBest hyperparameters: {'arch': 'resnet50', 'wd': 5.527401639726816e-05, 'epochs': 7, 'bs': 64, 'drop': 0.4}\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# Randomized search for hyperparameter optimization - Run 2\n\n\nFor n_trails = 10, the accuracy score and best hyper-parameter are as follows:\n\n| Trial | Best Score | Architecture | Weight Decay | Epochs | Batch Size | Drop |\n|-------|------------|--------------|--------------|--------|------------|------|\n| 0     | 0.9177     | ResNet34     | 0.000248     | 6      | 32         | 0.2  |\n| 1     | 0.9492     | ResNet50     | 0.002137     | 7      | 64         | 0.4  |\n| 2     | 0.9518     | ResNet50     | 0.000004     | 8      | 64         | 0.2  |\n| 3     | 0.9046     | ResNet34     | 0.000269     | 6      | 64         | 0.2  |\n| 4     | 0.9378     | ResNet50     | 0.000058     | 6      | 32         | 0.2  |\n| 5     | 0.9352     | ResNet18     | 0.000154     | 7      | 32         | 0.2  |\n| 6     | 0.9063     | ResNet34     | 0.000006     | 5      | 64         | 0.4  |\n| 7     | 0.9553     | ResNet50     | 0.000004     | 13     | 64         | 0.2  |\n| 8     | 0.9238     | ResNet34     | 0.000824     | 13     | 64         | 0.4  |\n| 9     | 0.9361     | ResNet18     | 0.000018     | 8      | 32         | 0.2  |\n\n\n**The best aacuracy score is 0.9553 with these hyperparameters:**\n- Architecture: ResNet 50\n- Weight Decay: 4e-6\n- Epochs: 5\n- Batch Size: 64\n- Drop: 0.2","metadata":{}},{"cell_type":"code","source":"import optuna\nfrom fastai.vision.all import *\nfrom sklearn.model_selection import cross_val_score\nfrom optuna.samplers import TPESampler\n\ndef objective(trial):\n    path = Path('/kaggle/input/brain-tumor-mri-classification-dataset/Brain_Tumor_MRI_Image_Dataset/Training')\n    \n    arch = trial.suggest_categorical('arch', ['resnet18', 'resnet34', 'resnet50'])\n    wd = trial.suggest_loguniform('wd', 1e-6, 1e-2)\n    epochs = trial.suggest_int('epochs', 5, 15)\n    bs = trial.suggest_categorical('bs', [32, 64])\n    drop = trial.suggest_categorical('drop', [0.2, 0.4])\n\n    dblock = DataBlock(blocks=(ImageBlock, CategoryBlock),\n                       get_items=get_image_files,\n                       get_y=parent_label,\n                       splitter=RandomSplitter(valid_pct=0.2, seed=42))\n    \n    dls = dblock.dataloaders(path, bs=bs)\n\n    learn = vision_learner(dls, arch, metrics=accuracy, wd=wd)\n    learn.fine_tune(epochs, base_lr=0.001, cbs=[MixedPrecision()])\n\n    return float(learn.validate()[1])\n\nsampler = TPESampler(seed=42)  # Initialize TPESampler for random search\nstudy = optuna.create_study(direction=\"maximize\", sampler=sampler)  # Use TPESampler for random search\nstudy.optimize(objective, n_trials=10)\n\ntrial = study.best_trial\n\nprint(\"Accuracy: {}\".format(trial.value))\nprint(\"Best hyperparameters: {}\".format(trial.params))","metadata":{"execution":{"iopub.status.busy":"2024-05-24T22:15:43.592939Z","iopub.execute_input":"2024-05-24T22:15:43.593255Z","iopub.status.idle":"2024-05-24T22:45:05.294958Z","shell.execute_reply.started":"2024-05-24T22:15:43.593221Z","shell.execute_reply":"2024-05-24T22:45:05.293995Z"},"trusted":true},"execution_count":6,"outputs":[{"name":"stderr","text":"[I 2024-05-24 22:15:43,603] A new study created in memory with name: no-name-b9e79e9e-2be7-4f61-82ff-92e3b89629f2\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.861309</td>\n      <td>0.381287</td>\n      <td>0.870403</td>\n      <td>00:13</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.560308</td>\n      <td>0.326959</td>\n      <td>0.875657</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.489167</td>\n      <td>0.303019</td>\n      <td>0.886165</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.382597</td>\n      <td>0.276688</td>\n      <td>0.898424</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.365501</td>\n      <td>0.247717</td>\n      <td>0.904553</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.277556</td>\n      <td>0.245371</td>\n      <td>0.908932</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.282006</td>\n      <td>0.237996</td>\n      <td>0.915937</td>\n      <td>00:17</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:17:45,930] Trial 0 finished with value: 0.917688250541687 and parameters: {'arch': 'resnet34', 'wd': 0.0002481040974867811, 'epochs': 6, 'bs': 32, 'drop': 0.2}. Best is trial 0 with value: 0.917688250541687.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.788204</td>\n      <td>0.412232</td>\n      <td>0.862522</td>\n      <td>00:21</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.351873</td>\n      <td>0.262150</td>\n      <td>0.925569</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.299193</td>\n      <td>0.231609</td>\n      <td>0.933450</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.243308</td>\n      <td>0.262405</td>\n      <td>0.929072</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.191433</td>\n      <td>0.175836</td>\n      <td>0.944834</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.133285</td>\n      <td>0.168542</td>\n      <td>0.948336</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.105654</td>\n      <td>0.230540</td>\n      <td>0.949212</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.096123</td>\n      <td>0.158684</td>\n      <td>0.953590</td>\n      <td>00:26</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:21:15,615] Trial 1 finished with value: 0.9492118954658508 and parameters: {'arch': 'resnet50', 'wd': 0.002136832907235877, 'epochs': 7, 'bs': 64, 'drop': 0.4}. Best is trial 1 with value: 0.9492118954658508.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.823435</td>\n      <td>0.355779</td>\n      <td>0.880035</td>\n      <td>00:21</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.334317</td>\n      <td>0.203773</td>\n      <td>0.931699</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.280687</td>\n      <td>0.215644</td>\n      <td>0.921191</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.238456</td>\n      <td>0.439207</td>\n      <td>0.925569</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.182599</td>\n      <td>0.181663</td>\n      <td>0.937828</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.150487</td>\n      <td>0.165346</td>\n      <td>0.942207</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.111429</td>\n      <td>0.183122</td>\n      <td>0.948336</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.089262</td>\n      <td>0.157622</td>\n      <td>0.952715</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.090734</td>\n      <td>0.156967</td>\n      <td>0.953590</td>\n      <td>00:25</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:25:07,888] Trial 2 finished with value: 0.9518388509750366 and parameters: {'arch': 'resnet50', 'wd': 3.6138942712165278e-06, 'epochs': 8, 'bs': 64, 'drop': 0.2}. Best is trial 2 with value: 0.9518388509750366.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>1.102986</td>\n      <td>0.427587</td>\n      <td>0.853765</td>\n      <td>00:12</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.541400</td>\n      <td>0.328082</td>\n      <td>0.880911</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.483583</td>\n      <td>0.297755</td>\n      <td>0.885289</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.401491</td>\n      <td>0.263839</td>\n      <td>0.893170</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.330477</td>\n      <td>0.252028</td>\n      <td>0.903678</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.283069</td>\n      <td>0.249856</td>\n      <td>0.908932</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.260506</td>\n      <td>0.250313</td>\n      <td>0.911559</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:26:59,027] Trial 3 finished with value: 0.9045534133911133 and parameters: {'arch': 'resnet34', 'wd': 0.000269264691008618, 'epochs': 6, 'bs': 64, 'drop': 0.2}. Best is trial 2 with value: 0.9518388509750366.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.667771</td>\n      <td>0.379004</td>\n      <td>0.904553</td>\n      <td>00:22</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.373117</td>\n      <td>0.274419</td>\n      <td>0.920315</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.327938</td>\n      <td>0.585635</td>\n      <td>0.930823</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.262243</td>\n      <td>0.608685</td>\n      <td>0.923818</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.216059</td>\n      <td>0.221038</td>\n      <td>0.944834</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.167353</td>\n      <td>4.164350</td>\n      <td>0.933450</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.139341</td>\n      <td>0.361042</td>\n      <td>0.941331</td>\n      <td>00:27</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:30:13,466] Trial 4 finished with value: 0.9378283619880676 and parameters: {'arch': 'resnet50', 'wd': 5.762487216478604e-05, 'epochs': 6, 'bs': 32, 'drop': 0.2}. Best is trial 2 with value: 0.9518388509750366.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.751824</td>\n      <td>0.272709</td>\n      <td>0.906305</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.453656</td>\n      <td>0.238539</td>\n      <td>0.918564</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.366793</td>\n      <td>0.219829</td>\n      <td>0.918564</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.302120</td>\n      <td>0.191546</td>\n      <td>0.925569</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.285857</td>\n      <td>0.202364</td>\n      <td>0.924694</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.228826</td>\n      <td>0.183159</td>\n      <td>0.929072</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.196506</td>\n      <td>0.176833</td>\n      <td>0.937828</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.171231</td>\n      <td>0.180764</td>\n      <td>0.934326</td>\n      <td>00:11</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:31:44,996] Trial 5 finished with value: 0.9352014064788818 and parameters: {'arch': 'resnet18', 'wd': 0.00015375920235481777, 'epochs': 7, 'bs': 32, 'drop': 0.2}. Best is trial 2 with value: 0.9518388509750366.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>1.101960</td>\n      <td>0.398535</td>\n      <td>0.851138</td>\n      <td>00:12</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.530839</td>\n      <td>0.325091</td>\n      <td>0.887916</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.482721</td>\n      <td>0.300519</td>\n      <td>0.889667</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.395030</td>\n      <td>0.281887</td>\n      <td>0.898424</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.338371</td>\n      <td>0.259621</td>\n      <td>0.905429</td>\n      <td>00:15</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.298018</td>\n      <td>0.260090</td>\n      <td>0.905429</td>\n      <td>00:15</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:33:18,873] Trial 6 finished with value: 0.9063047170639038 and parameters: {'arch': 'resnet34', 'wd': 6.0803901902966035e-06, 'epochs': 5, 'bs': 64, 'drop': 0.4}. Best is trial 2 with value: 0.9518388509750366.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.772641</td>\n      <td>0.359026</td>\n      <td>0.883538</td>\n      <td>00:20</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.361652</td>\n      <td>0.214777</td>\n      <td>0.930823</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.274222</td>\n      <td>0.192497</td>\n      <td>0.942207</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.240062</td>\n      <td>0.185538</td>\n      <td>0.932574</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.205064</td>\n      <td>0.181526</td>\n      <td>0.938704</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.175140</td>\n      <td>0.360629</td>\n      <td>0.940455</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.142931</td>\n      <td>0.150462</td>\n      <td>0.947461</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.112472</td>\n      <td>0.153896</td>\n      <td>0.949212</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.101038</td>\n      <td>0.140357</td>\n      <td>0.951839</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.081824</td>\n      <td>0.135689</td>\n      <td>0.952715</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.070126</td>\n      <td>0.148050</td>\n      <td>0.948336</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.072863</td>\n      <td>0.138243</td>\n      <td>0.953590</td>\n      <td>00:25</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.061516</td>\n      <td>0.130411</td>\n      <td>0.951839</td>\n      <td>00:26</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.053710</td>\n      <td>0.136767</td>\n      <td>0.955342</td>\n      <td>00:26</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:39:23,135] Trial 7 finished with value: 0.9553415179252625 and parameters: {'arch': 'resnet50', 'wd': 3.6618192203924288e-06, 'epochs': 13, 'bs': 64, 'drop': 0.2}. Best is trial 7 with value: 0.9553415179252625.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>1.072078</td>\n      <td>0.420000</td>\n      <td>0.844133</td>\n      <td>00:13</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.544761</td>\n      <td>0.344332</td>\n      <td>0.871278</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.517602</td>\n      <td>0.307067</td>\n      <td>0.890543</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.436178</td>\n      <td>0.276744</td>\n      <td>0.901051</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.380206</td>\n      <td>0.267414</td>\n      <td>0.901051</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.334762</td>\n      <td>0.257378</td>\n      <td>0.901926</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.268780</td>\n      <td>0.237314</td>\n      <td>0.919440</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.241955</td>\n      <td>0.232983</td>\n      <td>0.911559</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.218444</td>\n      <td>0.223770</td>\n      <td>0.917688</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.195902</td>\n      <td>0.217740</td>\n      <td>0.919440</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.184507</td>\n      <td>0.222164</td>\n      <td>0.914186</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.167970</td>\n      <td>0.211332</td>\n      <td>0.922942</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.157684</td>\n      <td>0.209978</td>\n      <td>0.924694</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.156113</td>\n      <td>0.210642</td>\n      <td>0.925569</td>\n      <td>00:16</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:43:13,933] Trial 8 finished with value: 0.9238178730010986 and parameters: {'arch': 'resnet34', 'wd': 0.0008241925264876454, 'epochs': 13, 'bs': 64, 'drop': 0.4}. Best is trial 7 with value: 0.9553415179252625.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.754382</td>\n      <td>0.306970</td>\n      <td>0.893170</td>\n      <td>00:10</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.449796</td>\n      <td>0.255285</td>\n      <td>0.911559</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.400212</td>\n      <td>0.207009</td>\n      <td>0.919440</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.325846</td>\n      <td>0.204704</td>\n      <td>0.919440</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.275052</td>\n      <td>0.171263</td>\n      <td>0.931699</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.235291</td>\n      <td>0.169300</td>\n      <td>0.935201</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.199903</td>\n      <td>0.156644</td>\n      <td>0.941331</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.197643</td>\n      <td>0.150987</td>\n      <td>0.939580</td>\n      <td>00:12</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.164178</td>\n      <td>0.148437</td>\n      <td>0.936077</td>\n      <td>00:12</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 22:45:05,289] Trial 9 finished with value: 0.9360770583152771 and parameters: {'arch': 'resnet18', 'wd': 1.753594952976443e-05, 'epochs': 8, 'bs': 32, 'drop': 0.2}. Best is trial 7 with value: 0.9553415179252625.\n","output_type":"stream"},{"name":"stdout","text":"Accuracy: 0.9553415179252625\nBest hyperparameters: {'arch': 'resnet50', 'wd': 3.6618192203924288e-06, 'epochs': 13, 'bs': 64, 'drop': 0.2}\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# Hyperparameter Optimization with Optuna's Successive Halving Pruner\n\nThe addition of SuccessiveHalvingPruner helps to improve efficiency during optimization. By early stopping trials with low validation accuracy (potentially bad performers), SHP focuses resources on trials that are more likely to be good. This can significantly reduce the total training time, especially for computationally expensive models.\n\nFor n_trails = 10, accuracy scores and hyper-paramters are:\n\n| Trial | Best Score | Architecture | Weight Decay | Epochs | Batch Size | Drop               |\n|-------|------------|--------------|--------------|--------|------------|--------------------|\n| 0     | 0.9623     | resnet50     | 2.057e-06    | 8      | 32         | 0.2888             |\n| 1     | 0.9518     | resnet50     | 0.001421     | 7      | 32         | 0.2707             |\n| 2     | 0.9807     | resnet34     | 3.744e-06    | 9      | 32         | 0.3918             |\n| 3     | 0.9641     | resnet18     | 2.667e-05    | 7      | 64         | 0.3196             |\n| 4     | 0.9711     | resnet34     | 0.005883     | 8      | 64         | 0.2000             |\n| 5     | 0.9650     | resnet18     | 5.694e-06    | 6      | 64         | 0.2266             |\n| 6     | 0.9737     | resnet18     | 1.813e-05    | 6      | 64         | 0.3732             |\n| 7     | 0.9851     | resnet34     | 0.004016     | 13     | 32         | 0.2680             |\n| 8     | 0.9667     | resnet50     | 0.008095     | 15     | 32         | 0.3013             |\n| 9     | 0.9632     | resnet50     | 0.0006995    | 6      | 32         | 0.3595             |\n\n\n**The best aacuracy score is 0.9851 with these hyperparameters:**\n\n- Architecture: ResNet 34\n- Weight Decay: 0.004016\n- Epochs: 13\n- Batch Size: 32\n- Drop: 0.268\n\nSHA identifies and eliminates underperforming configurations early on. This prevents further resource expenditure on trials unlikely to yield good results. Optuna's Successive Halving (SHA) pruner can be advantageous over a simple random search for hyperparameter optimization.","metadata":{}},{"cell_type":"code","source":"from fastai.vision.all import *\nimport optuna\nimport numpy as np\nfrom optuna.pruners import SuccessiveHalvingPruner\n\n# objective function for Optuna\ndef objective(trial):\n    path = Path('/kaggle/input/brain-tumor-mri-classification-dataset/Brain_Tumor_MRI_Image_Dataset/Training')\n\n    # hyperparameters\n    arch = trial.suggest_categorical('arch', [resnet18, resnet34, resnet50])\n    wd = trial.suggest_loguniform('wd', 1e-6, 1e-2)\n    epochs = trial.suggest_int('epochs', 5, 15)\n    bs = trial.suggest_categorical('bs', [32, 64])\n    drop = trial.suggest_float('drop', 0.2, 0.4)\n\n    #DataBlock\n    dblock = DataBlock(blocks=(ImageBlock, CategoryBlock),\n                       get_items=get_image_files,\n                       get_y=parent_label,\n                       splitter=RandomSplitter(valid_pct=0.2, seed=42))\n    \n    dls = dblock.dataloaders(path, bs=bs)\n    \n    # learner\n    learn = vision_learner(dls, arch, metrics=[accuracy], wd=wd)\n    \n    # Train the model\n    learn.fine_tune(epochs, base_lr=0.001, cbs=[MixedPrecision()])\n    \n    # validation accuracy\n    accuracy_metric = float(learn.validate()[1])\n    \n    return accuracy_metric\n\n# Optuna study with the Halving Pruner\nstudy = optuna.create_study(direction='maximize', pruner=SuccessiveHalvingPruner())\n\n# Optimize the objective function\nstudy.optimize(objective, n_trials=10)\n\n# best trial\ntrial = study.best_trial\n\nprint(\"Best Accuracy: {}\".format(trial.value))\nprint(\"Best hyperparameters: {}\".format(trial.params))","metadata":{"execution":{"iopub.status.busy":"2024-05-24T23:18:32.825211Z","iopub.execute_input":"2024-05-24T23:18:32.825891Z","iopub.status.idle":"2024-05-24T23:51:04.897150Z","shell.execute_reply.started":"2024-05-24T23:18:32.825856Z","shell.execute_reply":"2024-05-24T23:51:04.896207Z"},"trusted":true},"execution_count":8,"outputs":[{"name":"stderr","text":"[I 2024-05-24 23:18:32,835] A new study created in memory with name: no-name-3b9d03b9-fb10-4d8b-834d-1b4f896e975e\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.599265</td>\n      <td>0.300957</td>\n      <td>0.899300</td>\n      <td>00:23</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.264046</td>\n      <td>0.187430</td>\n      <td>0.922942</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.177461</td>\n      <td>0.192091</td>\n      <td>0.932574</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.111791</td>\n      <td>0.161029</td>\n      <td>0.949212</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.053418</td>\n      <td>0.128679</td>\n      <td>0.950088</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.039403</td>\n      <td>0.142595</td>\n      <td>0.951839</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.016710</td>\n      <td>0.131251</td>\n      <td>0.958844</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.020159</td>\n      <td>0.112070</td>\n      <td>0.965849</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.008556</td>\n      <td>0.110309</td>\n      <td>0.963222</td>\n      <td>00:28</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:22:49,077] Trial 0 finished with value: 0.9623467326164246 and parameters: {'arch': <function resnet50 at 0x7a55d092f880>, 'wd': 2.056715875355261e-06, 'epochs': 8, 'bs': 32, 'drop': 0.2887557735470722}. Best is trial 0 with value: 0.9623467326164246.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.638909</td>\n      <td>0.292574</td>\n      <td>0.915061</td>\n      <td>00:22</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.245941</td>\n      <td>0.205356</td>\n      <td>0.936953</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.183602</td>\n      <td>0.176353</td>\n      <td>0.943082</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.093307</td>\n      <td>0.156542</td>\n      <td>0.956217</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.046877</td>\n      <td>0.133005</td>\n      <td>0.960595</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.024073</td>\n      <td>0.149629</td>\n      <td>0.955342</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.021273</td>\n      <td>0.133934</td>\n      <td>0.957093</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.011730</td>\n      <td>0.134178</td>\n      <td>0.957093</td>\n      <td>00:28</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:26:35,485] Trial 1 finished with value: 0.9518388509750366 and parameters: {'arch': <function resnet50 at 0x7a55d092f880>, 'wd': 0.001421022419179072, 'epochs': 7, 'bs': 32, 'drop': 0.27067631749319276}. Best is trial 0 with value: 0.9623467326164246.\nDownloading: \"https://download.pytorch.org/models/resnet34-b627a593.pth\" to /root/.cache/torch/hub/checkpoints/resnet34-b627a593.pth\n100%|██████████| 83.3M/83.3M [00:01<00:00, 82.3MB/s]\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.707973</td>\n      <td>0.257959</td>\n      <td>0.910683</td>\n      <td>00:14</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.330547</td>\n      <td>0.151822</td>\n      <td>0.951839</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.178980</td>\n      <td>0.119798</td>\n      <td>0.951839</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.111872</td>\n      <td>0.113920</td>\n      <td>0.960595</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.064642</td>\n      <td>0.081061</td>\n      <td>0.976357</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.051511</td>\n      <td>0.100059</td>\n      <td>0.971979</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.026336</td>\n      <td>0.060369</td>\n      <td>0.983362</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.009389</td>\n      <td>0.068532</td>\n      <td>0.979860</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.007585</td>\n      <td>0.069661</td>\n      <td>0.978984</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.004051</td>\n      <td>0.067763</td>\n      <td>0.980736</td>\n      <td>00:17</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:29:36,291] Trial 2 finished with value: 0.9807355403900146 and parameters: {'arch': <function resnet34 at 0x7a55d092f6d0>, 'wd': 3.743608702148963e-06, 'epochs': 9, 'bs': 32, 'drop': 0.391776451145661}. Best is trial 2 with value: 0.9807355403900146.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.932631</td>\n      <td>0.280745</td>\n      <td>0.909807</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.344489</td>\n      <td>0.164103</td>\n      <td>0.943082</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.204429</td>\n      <td>0.138138</td>\n      <td>0.943958</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.107068</td>\n      <td>0.112685</td>\n      <td>0.964098</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.048349</td>\n      <td>0.085158</td>\n      <td>0.963222</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.023066</td>\n      <td>0.088834</td>\n      <td>0.962347</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.014993</td>\n      <td>0.089301</td>\n      <td>0.964098</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.008996</td>\n      <td>0.093573</td>\n      <td>0.962347</td>\n      <td>00:11</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:31:07,263] Trial 3 finished with value: 0.9640980958938599 and parameters: {'arch': <function resnet18 at 0x7a55d092f520>, 'wd': 2.666590746016712e-05, 'epochs': 7, 'bs': 64, 'drop': 0.3196090658228747}. Best is trial 2 with value: 0.9807355403900146.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.901441</td>\n      <td>0.251823</td>\n      <td>0.907180</td>\n      <td>00:12</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.349033</td>\n      <td>0.162082</td>\n      <td>0.948336</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.192588</td>\n      <td>0.135106</td>\n      <td>0.956217</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.092922</td>\n      <td>0.108112</td>\n      <td>0.963222</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.056528</td>\n      <td>0.101595</td>\n      <td>0.968476</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.029500</td>\n      <td>0.098100</td>\n      <td>0.972855</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.011884</td>\n      <td>0.093614</td>\n      <td>0.972855</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.005482</td>\n      <td>0.095259</td>\n      <td>0.970228</td>\n      <td>00:16</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.003129</td>\n      <td>0.106637</td>\n      <td>0.971103</td>\n      <td>00:16</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:33:36,021] Trial 4 finished with value: 0.971103310585022 and parameters: {'arch': <function resnet34 at 0x7a55d092f6d0>, 'wd': 0.005882789424928583, 'epochs': 8, 'bs': 64, 'drop': 0.20001891036926267}. Best is trial 2 with value: 0.9807355403900146.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.909503</td>\n      <td>0.293536</td>\n      <td>0.904553</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.358840</td>\n      <td>0.165944</td>\n      <td>0.937828</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.204989</td>\n      <td>0.119504</td>\n      <td>0.957968</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.092436</td>\n      <td>0.118193</td>\n      <td>0.962347</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.041803</td>\n      <td>0.119817</td>\n      <td>0.959720</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.024639</td>\n      <td>0.108614</td>\n      <td>0.964098</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.012232</td>\n      <td>0.106503</td>\n      <td>0.965849</td>\n      <td>00:11</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:34:56,095] Trial 5 finished with value: 0.9649737477302551 and parameters: {'arch': <function resnet18 at 0x7a55d092f520>, 'wd': 5.693699927575653e-06, 'epochs': 6, 'bs': 64, 'drop': 0.2265821977897189}. Best is trial 2 with value: 0.9807355403900146.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.911686</td>\n      <td>0.281321</td>\n      <td>0.899300</td>\n      <td>00:09</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.361514</td>\n      <td>0.165636</td>\n      <td>0.945709</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.184928</td>\n      <td>0.141123</td>\n      <td>0.953590</td>\n      <td>00:11</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.086210</td>\n      <td>0.109074</td>\n      <td>0.960595</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.043293</td>\n      <td>0.094568</td>\n      <td>0.970228</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.023215</td>\n      <td>0.077831</td>\n      <td>0.972855</td>\n      <td>00:10</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.011812</td>\n      <td>0.077520</td>\n      <td>0.974606</td>\n      <td>00:10</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:36:13,018] Trial 6 finished with value: 0.9737303256988525 and parameters: {'arch': <function resnet18 at 0x7a55d092f520>, 'wd': 1.812866118696492e-05, 'epochs': 6, 'bs': 64, 'drop': 0.37324777487924343}. Best is trial 2 with value: 0.9807355403900146.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.721485</td>\n      <td>0.267010</td>\n      <td>0.913310</td>\n      <td>00:13</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.305417</td>\n      <td>0.163025</td>\n      <td>0.943958</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.174053</td>\n      <td>0.099729</td>\n      <td>0.964098</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.104068</td>\n      <td>0.135178</td>\n      <td>0.956217</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.079441</td>\n      <td>0.131653</td>\n      <td>0.960595</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.077399</td>\n      <td>0.149306</td>\n      <td>0.961471</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.063717</td>\n      <td>0.177402</td>\n      <td>0.964974</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.031494</td>\n      <td>0.084892</td>\n      <td>0.979860</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.033331</td>\n      <td>0.077717</td>\n      <td>0.973730</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.018448</td>\n      <td>0.069668</td>\n      <td>0.983362</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.021099</td>\n      <td>0.061960</td>\n      <td>0.982487</td>\n      <td>00:17</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.011762</td>\n      <td>0.051196</td>\n      <td>0.985990</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.006565</td>\n      <td>0.047528</td>\n      <td>0.985990</td>\n      <td>00:18</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.006825</td>\n      <td>0.047503</td>\n      <td>0.985114</td>\n      <td>00:17</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:40:21,627] Trial 7 finished with value: 0.9851138591766357 and parameters: {'arch': <function resnet34 at 0x7a55d092f6d0>, 'wd': 0.004016134981130656, 'epochs': 13, 'bs': 32, 'drop': 0.2679709587877348}. Best is trial 7 with value: 0.9851138591766357.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.620788</td>\n      <td>0.288412</td>\n      <td>0.895797</td>\n      <td>00:22</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.301431</td>\n      <td>0.195068</td>\n      <td>0.933450</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.145843</td>\n      <td>0.167635</td>\n      <td>0.943958</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.085029</td>\n      <td>0.167764</td>\n      <td>0.947461</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.083989</td>\n      <td>0.156370</td>\n      <td>0.952715</td>\n      <td>00:28</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.054462</td>\n      <td>0.123673</td>\n      <td>0.958844</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.034490</td>\n      <td>0.124303</td>\n      <td>0.966725</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>6</td>\n      <td>0.034632</td>\n      <td>0.127395</td>\n      <td>0.960595</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>7</td>\n      <td>0.023373</td>\n      <td>0.122620</td>\n      <td>0.965849</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>8</td>\n      <td>0.017231</td>\n      <td>0.114999</td>\n      <td>0.971103</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>9</td>\n      <td>0.012671</td>\n      <td>0.110276</td>\n      <td>0.970228</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>10</td>\n      <td>0.011670</td>\n      <td>0.111809</td>\n      <td>0.968476</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>11</td>\n      <td>0.006216</td>\n      <td>0.096242</td>\n      <td>0.971103</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>12</td>\n      <td>0.003732</td>\n      <td>0.091567</td>\n      <td>0.968476</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>13</td>\n      <td>0.005209</td>\n      <td>0.098448</td>\n      <td>0.971979</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>14</td>\n      <td>0.001999</td>\n      <td>0.104521</td>\n      <td>0.969352</td>\n      <td>00:27</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:47:49,546] Trial 8 finished with value: 0.9667250514030457 and parameters: {'arch': <function resnet50 at 0x7a55d092f880>, 'wd': 0.008095083336439402, 'epochs': 15, 'bs': 32, 'drop': 0.30129588154296516}. Best is trial 7 with value: 0.9851138591766357.\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.643114</td>\n      <td>0.311658</td>\n      <td>0.901051</td>\n      <td>00:22</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: left;\">\n      <th>epoch</th>\n      <th>train_loss</th>\n      <th>valid_loss</th>\n      <th>accuracy</th>\n      <th>time</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <td>0</td>\n      <td>0.260481</td>\n      <td>0.190513</td>\n      <td>0.932574</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>1</td>\n      <td>0.164268</td>\n      <td>0.152902</td>\n      <td>0.950963</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>2</td>\n      <td>0.076222</td>\n      <td>0.153643</td>\n      <td>0.953590</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>3</td>\n      <td>0.045928</td>\n      <td>0.119837</td>\n      <td>0.966725</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>4</td>\n      <td>0.019451</td>\n      <td>0.116562</td>\n      <td>0.965849</td>\n      <td>00:27</td>\n    </tr>\n    <tr>\n      <td>5</td>\n      <td>0.019660</td>\n      <td>0.117123</td>\n      <td>0.964098</td>\n      <td>00:27</td>\n    </tr>\n  </tbody>\n</table>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":"\n<style>\n    /* Turns off some styling */\n    progress {\n        /* gets rid of default border in Firefox and Opera. */\n        border: none;\n        /* Needs to be in here for Safari polyfill so background images work as expected. */\n        background-size: auto;\n    }\n    progress:not([value]), progress:not([value])::-webkit-progress-bar {\n        background: repeating-linear-gradient(45deg, #7e7e7e, #7e7e7e 10px, #5c5c5c 10px, #5c5c5c 20px);\n    }\n    .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n        background: #F44336;\n    }\n</style>\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<IPython.core.display.HTML object>","text/html":""},"metadata":{}},{"name":"stderr","text":"[I 2024-05-24 23:51:04,891] Trial 9 finished with value: 0.9632224440574646 and parameters: {'arch': <function resnet50 at 0x7a55d092f880>, 'wd': 0.0006994648882504557, 'epochs': 6, 'bs': 32, 'drop': 0.3595360164107316}. Best is trial 7 with value: 0.9851138591766357.\n","output_type":"stream"},{"name":"stdout","text":"Best Accuracy: 0.9851138591766357\nBest hyperparameters: {'arch': <function resnet34 at 0x7a55d092f6d0>, 'wd': 0.004016134981130656, 'epochs': 13, 'bs': 32, 'drop': 0.2679709587877348}\n","output_type":"stream"}]}]}
\ No newline at end of file

From 5a33c9a268bb8e0a1df05ece008439f6b6570124 Mon Sep 17 00:00:00 2001
From: theiturhs <96874023+theiturhs@users.noreply.github.com>
Date: Sat, 25 May 2024 13:08:56 +0530
Subject: [PATCH 2/3] Update Cross Validation Techniques.md

---
 .../Cross Validation Techniques.md                          | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/Brain Tumor MRI Classification/Cross Validation Techniques.md b/Brain Tumor MRI Classification/Cross Validation Techniques.md
index 43fd7a08..065e0b87 100644
--- a/Brain Tumor MRI Classification/Cross Validation Techniques.md	
+++ b/Brain Tumor MRI Classification/Cross Validation Techniques.md	
@@ -1,7 +1,7 @@
 # Cross Validation Techniques for Brain Tumor MRI Classification
 ____
 
-The dataset on which the cross validation is carried out can be found [on kaggle](https://www.kaggle.com/datasets/theiturhs/brain-tumor-mri-classification-dataset/data). Find the implementation of this [in this notebook]().
+The dataset on which the cross validation is carried out can be found [on kaggle](https://www.kaggle.com/datasets/theiturhs/brain-tumor-mri-classification-dataset/data). Find the implementation of this CrossValidation_Techniques.ipynb notebook.
 
 ### Cross Validation Teachniques carried out are as follows:
 *Different techniques for distributing dataset into training and testing dataset*
@@ -19,10 +19,10 @@ The dataset on which the cross validation is carried out can be found [on kaggle
 | Repeated K-Fold CV	| 0.9585 |	0.9504 |	0.9838 |	0.9255 |	0.9703|
 | Stratified K-Fold CV	| 0.9548 |	0.9544	| 0.9838 |	0.9096 |	0.9667|
 
-**Out of these techniques, K-Fold CV technique gives better overall accuracy and class wise accuracy.**
+**Out of these techniques, K-Fold CV technique gives better overall accuracy and class wise accuracy. Therefore, this technique can be used to divide our dataset into training and testing set.**
 
 1. Hold Out Technique: Randomly the dataset was distributed into 80% training set and 20% validation set. The aacuracy achieved was 93.80%.
 2. K-Fold Cross Validation: This divides data into k equal-sized folds, trains the model k times, each time using k-1 folds as training data and one fold as validation data. The accuracy achieved was 96.04%.
 3. Repeated K-Fold: This repeats the k-fold CV process multiple times with different random splits of the data. It achieved 95.85% accuracy.
 4. Leave One Out: It is a special case of k-fold CV where k is equal to the number of samples in the dataset. Each sample is used as a validation set once. So this was not implemented. It was computationally expensive, would have required a lot of training time.
-5. Stratified K-Fold: It is like k-fold CV, but ensures that each fold preserves the percentage of samples for each class. It achieved an accuracy of 95.48%.
\ No newline at end of file
+5. Stratified K-Fold: It is like k-fold CV, but ensures that each fold preserves the percentage of samples for each class. It achieved an accuracy of 95.48%.

From 789904dcf20e5ee1c918abe2ac0c0c93dcd64501 Mon Sep 17 00:00:00 2001
From: theiturhs <96874023+theiturhs@users.noreply.github.com>
Date: Sat, 25 May 2024 13:14:14 +0530
Subject: [PATCH 3/3] Update Hyper-Parameter Tuning.md

---
 .../Hyper-Parameter Tuning.md                 | 31 ++++++++++++++++---
 1 file changed, 27 insertions(+), 4 deletions(-)

diff --git a/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md b/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md
index 41cc2603..9807760f 100644
--- a/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md	
+++ b/Brain Tumor MRI Classification/Hyper-Parameter Tuning.md	
@@ -1,7 +1,6 @@
 # Hyper-Parameter Tuning Techniques for Brain Tumor MRI Classification
-____
 
-The dataset on which the cross validation is carried out can be found [on kaggle](https://www.kaggle.com/datasets/theiturhs/brain-tumor-mri-classification-dataset/data). Find the implementation of this [in this notebook](). Learner module, FastAI, is considered to provide a convenient way to create and fine-tune convolutional neural network (CNN) models. vision.learner is a function that helps us to construct a learner object, which has the model architecture, data, training configuration, and other elements. We can specify a pre-trained model architecture and fine-tune it on the dataset.
+The dataset on which the cross validation is carried out can be found [on kaggle](https://www.kaggle.com/datasets/theiturhs/brain-tumor-mri-classification-dataset/data). Find the implementation of this in Hyper-parameter_Tuning.ipynb notebook. Learner module, FastAI, is considered to provide a convenient way to create and fine-tune convolutional neural network (CNN) models. vision.learner is a function that helps us to construct a learner object, which has the model architecture, data, training configuration, and other elements. We can specify a pre-trained model architecture and fine-tune it on the dataset.
 
 ### Hyper-Parameter Tuning Teachniques carried out are as follows:
 *Different techniques to find out suitable hyper-parameters*
@@ -13,7 +12,7 @@ The dataset on which the cross validation is carried out can be found [on kaggle
 | -- | -- | -- | -- | -- | -- | -- |
 | Random Search optimization algorithm - Run 1 | 95.27% | ResNet50 | 5.527e-5 | 7 | 64 | 0.4 |
 | Random Search optimization algorithm - Run 2 | 95.53% | ResNet50 | 4e-6 | 5 | 64 | 0.2 |
-| Hyperparameter Optimization with Optuna's Successive Halving Pruner | 
+| Hyperparameter Optimization with Optuna's Successive Halving Pruner | 98.51% | ResNet18 | 0.004016 | 13 | 32 | 0.2680 |
 
 **Out of these techniques,Hyperparameter Optimization with Optuna's Successive Halving Pruner technique gives better overall accuracy. It almost take 40-50 minutes for each to fine-tune the model.**
 
@@ -36,6 +35,14 @@ For n_trails = 10, the accuracy score and best hyper-parameter are as follows:
 | 8         | 0.9212     | ResNet18     | 0.0002       | 5      | 64         | 0.4  |
 | 9         | 0.9203     | ResNet34     | 0.0007       | 13     | 32         | 0.4  |
 
+The best accuracy score is 0.9527 with these hyperparameters:
+
+- Architecture: ResNet 50
+- Weight Decay: 5.527e-5
+- Epochs: 7
+- Batch Size: 64
+- Drop: 0.4
+
 #### Random Search Optimizing Algorithm - Run 2
 
 For n_trails = 10, the accuracy score and best hyper-parameter are as follows:
@@ -53,6 +60,14 @@ For n_trails = 10, the accuracy score and best hyper-parameter are as follows:
 | 8     | 0.9238     | ResNet34     | 0.000824     | 13     | 64         | 0.4  |
 | 9     | 0.9361     | ResNet18     | 0.000018     | 8      | 32         | 0.2  |
 
+The best accuracy score is 0.9553 with these hyperparameters:
+
+- Architecture: ResNet 50
+- Weight Decay: 4e-6
+- Epochs: 5
+- Batch Size: 64
+- Drop: 0.2
+
 #### Hyperparameter Optimization with Optuna's Successive Halving Pruner
 
 For n_trails = 10, accuracy scores and hyper-paramters are:
@@ -70,9 +85,17 @@ For n_trails = 10, accuracy scores and hyper-paramters are:
 | 8     | 0.9667     | resnet50     | 0.008095     | 15     | 32         | 0.3013             |
 | 9     | 0.9632     | resnet50     | 0.0006995    | 6      | 32         | 0.3595             |
 
+The best accuracy score is 0.9851 with these hyperparameters:
+
+- Architecture: ResNet 34
+- Weight Decay: 0.004016
+- Epochs: 13
+- Batch Size: 32
+- Drop: 0.268
+
 ##### Summarizing
 
 1. Random search optimization gives almost 95% accuracy when we ran it two times each with 10 iterations.
 2. Hyperparameter Optimization with Optuna's Successive Halving Pruner, we are getting 98.51% which is a remarkable accuracy.
 
-**So we will be using k-fold validation technique followed by Hyperparameter Optimization with Optuna's Successive Halving Pruner for getting the appropriate hyper-parameters.**
\ No newline at end of file
+**So we will be using k-fold validation technique followed by Hyperparameter Optimization with Optuna's Successive Halving Pruner for getting the appropriate hyper-parameters.**