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

Yihong Wu, Yale University, Spring 2024

The interplay between information theory and statistics is a constant theme in the development of both fields. This course will discuss how techniques rooted in information theory play a key role in understanding the fundamental limits of high-dimensional statistical problems in terms of minimax risk and sample complexity. In particular, we will rigorously justify the phenomena of dimensionality reduction by either intrinsic low-dimensionality (sparsity, smoothness, shape, etc) or - the less familiar - extrinsic low-dimensionality (functional estimation). Complementing this objective of understanding the fundamental limits, another significant direction is to develop computationally efficient procedures that attain the statistical optimality, or to understand the lack thereof.

  • Lectures: Tuesdays 4pm–550pm KT 219

  • Office hours: by appointment

  • Syllabus

  • We will draw materials from this book draft and this set of lecture notes. Handout for other materials will be provided.

Announcements

  • Welcome to S&DS677 Spring 2024. The first lecture is Jan 16.