1228: Probabilistic Graphical Models

This page was created for posting some useful material for the course Probabilistic Graphical Models developed and taught by Prof. Joschka Boedecker and me starting from 2024.

We have decided to rewrite some of the materials here from the summer of 2026, and to add more topics. So you might want to check this webpage occasionally to get updated.

Catalog description

The lecture deals with methods of Probabilistic Graphical Models that constitute an important class of machine learning algorithms. A brief content about the topics of the course can be found below.

  1. Introduction

  2. Bayesian classifiers

  3. Markov models

  4. Bayesian networks

  5. Inference with Monte Carlo methods

  6. Markov decision problems

  7. Control as probabilistic inference

Support notes

The old set of lecture notes, used until 2025, are available here.

Textbooks and references

There is no textbook. Everything we'll use will be posted on the this webpage in pdf format.

If you’d like to consult some books, we listed some below.

Prerequisites

There are no strict prerequisites, but good knowledge of probability would significantly reduce the learning curve. Exposure to graph theory, optimization, and reinforcement learning helpful but not required.

Final exam

The final exam is a 90 minutes open-book exam. You are allowed to bring one cheat sheet subject to the following requirements:

  • The cheat sheet could be (at most) one double sided A4 paper.

  • The content must be hand-written.

  • Yes, you could write digitally and then print, but a magnifier is not allowed in the exam.

No other materials (e.g., textbooks) are allowed in the exam. We will confirm if a calculator is strictly necessary closer to the exam date, but you can nevertheless bring one as you like.

Some old exams

The following final exams from the previous years can be used for practice.