S&DS677: Topics in High-Dimensional Statistics and Information Theory

Yihong Wu, Yale University, Spring 2024

Lecture schedule

Except otherwise noted, the reading materials below is from

topicsreadingnotes and references
Jan 16 Introduction (slides), basics of statistical decision theory Sec 28.1-28.3
Jan 23 Minimax theorem and duality, tensorization of experiments, Anderson's lemma Sec 28.4-28.6
Jan 30 f-divergence I: definitions, properties, examples, hypothesis testing Sec 7.1-7.3
Feb 6 f-divergence II: inequalties and joint range, variational representation, GAN, information bound Sec 7.4-7.5, 29.1-29.2
Feb 13 f_divergence III: chain rule and tensorization, Donsker-Varadhan and Gibbs Sec 4.3-4.4 Jayram's chainrule for squared Hellinger
Feb 20 f_divergence IV: PAC Bayes and applications, variational inference Sec 4.8, 7.13 Applications to smallest singular value: Oliveira and Moutada;
dim-free concentration on sample covariance matrix: Zhivotovskiy
Feb 27 Mutual information method Sec 30
Mar 5 Le Cam's method (two-point), Assoud's lemma (hypercube), Fano's method (packing) Sec 31
Mar 12 Spring break I 😄
Mar 19 Spring break II 😂
Mar 26 Metric entropy I Sec 27.1-2, 27.4
Apr 2 Metric entropy II Sec 27.3, 27.5
Apr 9 Entropic upper bound for statistical estimation (Yang-Barron, Le Cam-Birgé, Yatracos) Sec 32
Apr 16 functional estimation and HT I: goodness-of-fit testing, second moment method and truncation Lec 23 of these lecture notes See (3.7) and the surrounding discussion in this paper by Gerber-Polyanskiy for a tentalizing conjecture of the sample complexity of goodness of fit testing versus estimation and many examples
Apr 23 functional estimation and HT II: sparse detection, uniformity testing Lec 24 of these lecture notes Survey by Cannone