This repository contains Jupyter notebooks implementing the code samples found in the book Deep Learning with Python (Manning Pubblications) by François Chollet, gathered by chapter.

The author already published notebooks with the exercises in his Github account. Those you can find in this repository are made by myself for my own learning, with no intention so steal any intellectual property. Exercises that worked flawlessly in TF 2 have been simply copied.

• Create a virtual environment with conda or virtualenv. Recommended conda with Python 3.6.

conda create --name <your_project> python=3.6

• Install dependencies:

pip install -r requirements.txt

The exercises in the book are written for tensorflow 1.* and Keras 2.0.8. All the code in this repo have been rewritten to work with tensorflow 2.2.* and the corresponding Keras version 2.2.4-tf.

No relevant exercises

• 01: a first look at a neural network - notebook

• 01: binary classification - notebook
• 02: multiclass classification - notebook
• 03: logistic regression - notebook

• 01: binary classification: mitigate overfitting and underfitting - notebook

• 01: introduction to CNN - notebook
• 02: using CNNs with small datasets - notebook
• 03: using a pretrained CNN - notebook
• 04: visualizing what a CNN learn - notebook

• 01: one-hot encoding of words or characters - notebook
• 02: using word embeddings - notebook
• 03: understanding RNNs - notebook
• 04: advanced use of RNNs - notebook
• 05: sequence processing with convnets - notebook

• 01: the Keras functional API - notebook

• 01: text generation with LSTM - notebook
• 02: Deep Dream - notebook
• 03: neural style transfer - notebook
• 04: variational autoencoders (VAEs) - notebook
• 05: introduction to GANs - notebook