Convolutional Neural networks applied to CIFAR10 images Dataset

A deep neural network using the AI tensorflow API Keras and advanced Deep Learning Techniques.

Key Features • How To Use • Download • Credits • Related • License

Key Features

This deep learning model predicts the class of object into an image. This prediction is given in a string type. The dataset is taken from the CIFAR10 Dataset . So here are the key features of this project:

• Model is built using a Convolutional Neural Network Architecture with 1,188,170 trainable parameters. We use optimal methods such like
  • Pooling
  • Dropout
  • BatchNormalization
  • Regularizers
  • Padding
• We got a 85.2% of validation accuracy.
• The learning curve shows a small amount of overfit. The graph is the following

How To Use

To clone and run this application, you will need Git.

# Clone this repository
$ git clone https://github.com/santiagoahl/cifar10-classification.git

# Go into the repository
$ cd cifar10-classification

# Install Jupyter Notebooks
$ pip install jupyterlab
$jupyter-lab
$pip install notebook

# Run
$jupyter notebook

Credits

This model uses the following open source datasets and packages:

• Keras
• Tensorflow
• CIFAR10 Dataset
• Pandas
• Numpy
• Matplotlib

License

MIT

---

Web Site santiagoal.super.site  ·  GitHub @santiagoahl  ·  Twitter @sahumadaloz