Theoretical Machine Learning (Fall 2026)

  • Schedule is subject to small adjustments.

  • Lecture notes will be posted before each lecture (and might be updated slightly soon after the lecture).

Date Topics Lecture Notes and Recommended Reading
08/28 Introduction, statistical learning, online learning,
no free lunch theorem, online-to-batch conversion
Lecture notes 1
Sections 3-5 of R&S
09/04 Uniform convergence, Rademacher complexity, finite class
Classification: growth function, VC dimension, Sauer's lemma
Lecture notes 2
09/11 Regression: covering number, Dudley entropy integral,
chaining technique, Pseudo-dimension
Lecture notes 3
Sections 12.4 and 12.9 of R&S
09/18 Regression: fat-shattering dimension
Case study on neural nets: dimension-independent covering number
spectral complexity, margin
Lecture notes 4
Section 12.8 of R&S
Proof for bounding covering number with fat-shattering dimension
Zhang-Bengio-Hardt-Recht-Vinyals’17
Bartlett-Foster-Telgarsky’17
09/25 Online learning: empirical process with dependent data,
sequential Rademacher complexity
Lecture notes 5
Section 13.5 of R&S
10/02 Online learning with a finite class
Online classification: zero-covering number and Littlestone dimension
Online regression: covering number, chaining, fat-shattering dimension
Lecture notes 6
Section 14.5 of R&S
10/09 Fall Recess
10/16 Online algorithms for finite classes: Halving and Hedge
Online algorithms for infinite classes with bounded Littlestone dimension
Perceptron
Lecture notes 7
10/23 Online Convex Optimization, Follow-the-Regularized-Leader
From values to algorithms
Learning with partial information, multi-armed bandits,
EXP3, Explore-then-Exploit
Lecture notes 8
More on OCO and FTRL from this note
10/30 UCB, Lower bound for multi-armed bandits
Partial monitoring: classification theorem
11/06 Partial monitoring: algorithms and lower bounds
11/13 Student presentations
11/20 Student presentations
11/27 Thanksgiving
12/04 Student presentations