Bayesian Theory and Computation, Spring 2022

Basic Info

  • Instructor: Cheng Zhang (chengzhang@math.pku.edu.cn)
  • Teaching Assistant: Longlin Yu (1800010783@pku.edu.cn)
  • Class times and location: odd Wednesday 1:00-2:50pm, Friday 3:10-5:00pm, Classroom Building No.3, Room 105
  • Office hours: Thursday 3:00-5:00pm or by appointment, 315 Building No.20
  • Syllabus

Description and Objectives

The objective of this course is to explore Bayesian statistical theories and methods, and discuss their application in real life problems. Students would learn how to formulate a scientific question by constructing a Bayesian model, and perform Bayesian statistical inference to answer that question. Throughout this course, students would be exposed to the theory of Bayesian inference. They would also learn several computational techniques, such as importance sampling, sequential Monte Carlo, Markov Chain Mote Carlo (MCMC) algorithms, variational inference (VI), and use these techniques for Bayesian analysis of real data. Additional topics may vary. Coursework will include computer assignments.

Prerequisites

Multivariate calculus, linear algebra, graduate level courses in probability and statistics, applied stochastic processes

Main textbook

  • Gelman, A., Carlin, J., Stern, H., and Rubin, D. (2003). Bayesian Data Analysis, 2nd Edition, Chapman & Hall.

Other interesting references

  • Liu, J. (2001). Monte Carlo Strategies in Scientific Computing, Springer-Verlag.
  • Lange, K. (2002). Numerical Analysis for Statisticians, Springer-Verlag, 2nd Edition.
  • Keener, R. W. (2010). Theoretical Statistics: Topics for a Core Course, Springer.
  • Christian, P. R. (2004). The Bayesian Choice, Springer.

Grading

  • 4 problem sets (15% each), 60%
  • final project (40%): midterm proposal (5%) + oral presentation (10%) + final write-up (25%)

There will be 7 free late days in total, use them in your own ways. Afterwards, late homework will be discounted by 25% for each additional day. Not acceptable after 3 late days per problem set (PS). Late policy does not apply to the final project, please submit it on time. Discussing assignments verbally with your classmates is allowed and encouraged. However, you should finish your work independently. Identified cheating incidents will be reported and will result in zero grades.

Computer and Technical Requirements

We will use python during the course. A good Python tutorial is available at http://www.scipy-lectures.org/. You may also find another shorter tutorial useful at http://cs231n.github.io/python-numpy-tutorial/. If you have never used Python before, I recommend using Anaconda Python 3.7 https://www.continuum.io/.

Lectures

Assignments

Final Project

You may structure your project exploration around a general problem type, algorithm, or data set, but should explore around your problem, testing thoroughly or comparing to alternatives. You may work on the project as teams. Each team may have up to 4 people. Please form your team by the end the 4th week. You should present a project proposal that briefly describe your project concept and goals in one slide at midterm. You should turn in a write-up (< 10 pages) describing your project and its outcomes, similar to a research-level publication. I suggest the latex styles for NeurIPS or ICLR. There will be in class project presentation at the end of the term. Not presenting your projects will be taken as voluntarily giving up the opportunity for the final write-ups.

Tentative Schedule

Week Date Topics Notes
1 02/23 Introduction  
  02/25 Single- and Multi- Parameter Models  
2 03/04 Monte Carlo Methods, Markov Processes  
3 03/09 Metropolis Hastings, Gibbs Sampling  
  03/11 Parallel Tempering, Slice Sampling, Hierarchical Models  
4 03/18 Linear Models, Generalized Linear Models  
5 03/23 Bayesian Decision Theory, Bayesian Model Selection  
  03/25 Importance Sampling, Variance Reduction Techniques  
6 04/01 Sequential Monte Carlo  
7 04/06 MALA, Hamiltonian Monte Carlo, Adaptive MCMC  
  04/08 Scalable MCMC Methods  
8 04/15 Convergence Theory of MCMC  
9 04/20 Expectation Maximization  
  04/22 Variational Bayesian EM proposal presentation
10 04/29 Variational Inference, Mean Field VI  
11 05/04 – Labour Day
  05/06 – Labour Day
12 05/13 Stochastic Variational Inference  
13 05/18 Gaussian Processes, Sparse Gaussian Processes  
  05/20 Dirichlet Processes, DP Mixture Models  
14 05/27 Project Presentation  
15 06/01 Project Presentation  
  06/03 – Dragon Boat Festival
16 06/10 Project Presentation