File: cats_dogs_classification.ipynb
Objectives:
- Building powerful image classification models using very little data
- Using image data-augmentation techniques
- Predicting cats and dogs--Kaggle
- Calcualting model weights, stage-by-stage
- Saving model-weights and model-architecture to files
- Loading saved model-architecture and model-weights
File: cifar10_classification.ipynb
Objective:
- CIFAR10 image classification using deep learning
File: cifar10_classification_functionalModeling.ipynb
Objective:
- CIFAR10 image classification with deep-learning CNN
- Building Functional models (Use GPU not CPU)
File: classify_with_vgg16_softmax.ipynb
Objectives:
- Transfer Learning: Building powerful image classification models using very little data using pre-trained applications
- Feature Enngineering: Using engineered features with Random Forest Classifier
- Learning Rate Annealing: Moderating Learning rate on/near plateau
File: generators in python.ipynb
Objectives:
- How a generator works
- Applications of a generator
File: expts_with_mobilenet.ipynb
Objectives:
- Using mobilenet
- Using mobilenet on browser
- Using Image Augmentaion techniques from libraries other than tensorflow/keras
- Transfer learning using Mobilenet
File: keras_functional.ipynb
Objective:
- Using keras functional API
File: image_augmentation.ipynb
Objective(s):
How Image augmentation is performed. Image augmentation basics
File: learningRateScheduler_tensorbard.ipyb****
*Objectives: *
- To use multiple callbacks
- To use learning-rate scheduler
- To see results on tensorboard
File(s):
keras_hyperparameterOpt_class.ipynb
keras_hyperparameterOpt.ipynb
Objective:
- Hyperparameter tuning example of Neural Network using keras-tuner of dense network
File: plot_vgg16_layer_features.ipynb
Objectives:
- Experimenting with Very Deep ConvNets: VGG16
- Peeping into layers and plotting extracted-features
- Visualize filters
File: pretrained_layers_autoencoder_I.ipynb
Objectives
- Building autoencoder using Model class subclassing
- Using pre-trained autoencoder layers in a classifier
- Comparing Classifer performance with and without pre-trained
- Using keras model as a layer
- A pre-trained model using autoencoder gives better classification
File: pretrained_layers_autoencoder_II.ipynb
Objectives:
- Building autoencoder using Model class subclassing
- Training autoencoder with gaussian noise added
- Using pre-trained autoencoder layers in a classifier
- Comparing Classifer performance with and without pre-trained
- Using keras model as a layer
- A pre-trained model using autoencoder-with-noise added gives better classification
File: pretrained_layers_autoencoder_III.ipynb
Objectives
- Building autoencoder using Model class subclassing
- Building autoencoder using Functional API
- Training autoencoder with gaussian noise added
- Using pre-trained autoencoder layers in a classifier
- Comparing Classifer performance with and without pre-trained
- Using keras model as a layer
- A pre-trained model using autoencoder-with-noise added gives better classification
File: subClassingKerasModel.ipynb
Objective:
- Subclassing keras 'Model' class to create Dynamic models
- Two examples
File: reusing_trained_layers.ipynb
*Objective: *
- Reuse layers from one neural network in another.
- Ist model was trained using a different set of class labels than present in the IInd case