Lecture notes for UW-Madison STAT615, Spring 2026, organized into four chapters.

  1. Prediction and consistency — Bayes classifiers, local averaging, concentration inequalities, and consistency. Lectures 0-9.
  2. Empirical risk and generalization — ERM, uniform convergence, VC dimension, and Rademacher complexity. Lectures 9-13.
  3. Linear classifiers and support vector machines — LDA, QDA, logistic regression, margins, and duality. Lectures 13-17.
  4. Kernels, Gaussian processes, and conformal prediction — RKHS, kernel regression, spectral expansions, and prediction sets. Lectures 17-23.

Original lecture PDF