STAT520P: Bayesian Optimization

  • 24 Oct 2023 — 7 Dec 2023
  • Tuesdays and Thursdays, 1:00 — 2:30PM
  • Earth Science Building, Room #4192
  • (Office hours immediately follow Thursday class, ESB 3142.)


Subject to change; check announcements and Canvas!.

Class/Date Class Topic Reading Due Assignment Due
Class 1 (Tues, Oct 24) Introduction, course logistics
Class 2 (Thurs, Oct 26) Gaussian processes
(Cheat sheet on multivariate Gaussians)
Textbook Ch. 1-2
Class 3 (Tues, Oct 31) Gaussian processes and kernels
(Jupyter demo, David Duvenaud's kernel cookbook)
Textbook Ch. 3, 9.1 (optional) Diagonstic problem set (TeX template and math_commands.tex macros file)
Class 4 (Thurs, Nov 2) Decision theory and optimization theory Textbook Ch. 5
Class 5 (Tues, Nov 7) (Jupyter demo on BDT acquisition functions) Acquisition functions Textbook Ch. 6-7 Project proposal (for those choosing the project final asssessment option)
Class 6 (Thurs, Nov 9) (Jupyter demo on UCB, Jupyter demo on Thompson sampling ) Acquisition functions
No Class on Nov 14
(Reading Week)
Class 7 (Thurs, Nov 16) Implementation details Textbook Ch. 8, 9.2 Short coding assignment
Class 8 (Tues, Nov 21) Special Topic (Multi-Objective BO) Multi-Objective Bayesian Optimization over High-Dimensional Search Spaces
Class 9 (Thurs, Nov 23) Special Topic (Local BO) Scalable Global Optimization via Local Bayesian
Class 10 (Tues, Nov 28) Special Topic (Mixed-Space BO) Bayesian Optimization over Discrete and Mixed Spaces via Probabilistic Reparameterization
Class 11 (Thurs, Nov 30) Special Topic (Causal BO) Causal Bayesian Optimization
Class 12 (Tues, Dec 5) Special Topic (Preference BO) Bayesian Active Learning for Classification and Preference Learning
Class 13 (Thurs, Dec 7) Special Topic (Cost-Aware BO) Budgeted Bandit Problems with Continuous Random Costs