Course positioning
CS 229M (STATS 214) is the most mathematically demanding ML course, aimed at graduate students who want to do theory. It does not teach model tuning; it trains you to analyze modern large-scale nonlinear models with concentration inequalities, uniform convergence, and optimization theory.
What it covers
- Concentration inequalities: Hoeffding, Chernoff, McDiarmid
- Generalization theory: uniform convergence, VC dimension, Rademacher complexity
- Non-convex optimization: gradient descent, saddle points, landscape
- Implicit regularization: overparameterization, double descent, margin stability
- Unsupervised learning and domain adaptation: representation learning theory
- Modern model analysis: approximation and generalization of Transformers/deep networks
Prerequisites
MATH 51; STATS 117; CS 229 or STATS 315A.
Assessment & workload
Assignments plus a final/project, mostly proofs and paper reading. Check the current syllabus for details.
Who it's for
People who want to do ML theory, a PhD, or rigorous analysis; also suitable for math/statistics backgrounds moving into ML.
Study advice
Brush up on real analysis and probability first. Don't expect to understand it in one pass — work through the notes and papers repeatedly.
Related courses
Official course description
How do we use mathematical thinking to design better machine learning methods? This course focuses on developing mathematical tools for answering this question. This course will cover fundamental concepts and principled algorithms in machine learning, particularly those that are related to modern large-scale non-linear models. The topics include concentration inequalities, generalization bounds via uniform convergence, non-convex optimization, implicit regularization effect in deep learning, and unsupervised learning and domain adaptations. Prerequisites: MATH 51 and STATS 117 and either CS 229 or STATS 315A. See https://statistics.stanford.edu/course-equiv for equivalent courses in other departments that satisfy these prerequisites.