Mathematics Foundations

Four 3-hour sessions covering the linear algebra and calculus behind machine learning: vectors and cosine similarity, matrices as transformations and the forward pass, derivatives and the chain rule as backpropagation, and gradient descent.

The slides are made from my handwritten notes, taken live during class.

Sessions

• Session 1: vector addition and the dot product, cosine similarity on real data.
• Session 2: matrices as linear transformations, the forward pass, eigenvectors and PCA.
• Session 3: derivatives, differentiation rules, the chain rule as backpropagation.
• Session 4: the gradient, gradient descent, local and global minima.