1. Information Theory Background for Machine Learning 

2. Neural Networks Theory, non-linearity, learning through backpropagation and gradient descend 

3. PyTorch Introduction 

4. Building a feed-forward network from scratch with PyTorch

5. Overfitting and Underfitting a Neural Network for universal approximation. Dropout and regularizations.

An Advanced Example (6 hours)

1. Convolutional Neural Networks

2. Variational Auto-Encoder for image denoising 

3. (OR in alternatively) Generative Adversarial Networks

Sustainable AI: an example (4 hours)

1. Motivation for energy efficient deep learning 

2. Pruning Neural Networks and Lottery Ticket Hypothesis