Layer 0

Part 2 — Linear Algebra for ML

Models do not learn “meaning”. They learn geometry: vectors, distances, similarity, and transformations. This part defines the mathematical space where learning happens.

Purpose

Linear algebra is the language of representations. Every dataset becomes a matrix, every model is a function over vectors, and many learning algorithms are just repeated linear transforms plus non-linearities.

Concepts

Chapter 0.4 — Scalars, Vectors, and Feature Spaces Chapter 0.5 — Datasets as Matrices Chapter 0.6 — Dot Products, Projections, and Similarity Chapter 0.7 — Norms, Distances, and Geometry Chapter 0.8 — Linear Transformations and Model Layers Chapter 0.9 — Rank, Collinearity, and Numerical Stability Chapter 0.10 — Eigenvalues, PCA, and Dimensionality Reduction