Master's year 1 · 21 hours

Mathematics for ML and Deep Learning

The mathematical foundations of deep learning: linear algebra and SVD, multivariate calculus, optimization and information theory, with Python tutorials.

Learning objectives

• Compute gradients, Jacobians and Hessians, and formalize backpropagation
• Apply SVD and PCA to dimensionality reduction
• Understand convergence, saddle points and L1/L2 regularization
• Connect cross-entropy and KL divergence to the choice of loss function

Course outline Free preview

1. Linear algebra, SVD and PCA
2. Tutorial: SVD on images and PCA
3. Multivariate calculus
4. Tutorial: gradients of a two-layer network, autograd and JAX
5. Probability and information theory
6. Tutorial: SGD and Adam in practice