Optimisation for Artificial Intelligence, a 4-day course

View the Project on GitHub xoolive/optim4ai

Optimisation for Machine Learning

« Previous | Home ↑ | Next »

This section will bring the optimisation topics we addressed with gradient descent methods, to Machine Learning and the training of neural networks. To date, there are two most widespread ML libraries for Python, both based on similar concepts: TensorFlow and PyTorch.

We will focus here on PyTorch and look into how the library addresses optimisation with particular structures called tensors. We focus in the whole section on optimisation problems, and leave neural networks aside.

3.1 Introduction to tensors in PyTorch
3.2 Automatic differentation
3.3 Optimisers in PyTorch
3.4 Exercice: Linear regression

The notebook producing these pages is available on the GitHub page, but focus on reading here: blindly executing cells may only distract you from the main content.