Bayesian Machine Learning and Information Processing (5SSD0)

The course “Bayesian Machine Learning and Information Processing” (5SSD0) starts in November 2024 (Q2).

Course goals

This course provides an introduction to Bayesian machine learning and information processing systems. The Bayesian approach affords a unified and consistent treatment of many useful information processing systems.

Course summary

This course covers the fundamentals of a Bayesian (i.e., probabilistic) approach to machine learning and information processing systems. The Bayesian approach allows for a unified and consistent treatment of many model-based machine learning techniques. Initially, we focus on Linear Gaussian systems and will discuss many useful models and applications, including common regression and classification methods, Gaussian mixture models, hidden Markov models and Kalman filters. We will discuss important algorithms for parameter estimation in these models including the Variational Bayes method. The Bayesian method also provides tools for comparing the performance of different information processing systems by means of estimating the Bayesian evidence for each model. We will discuss several methods for approximating Bayesian evidence. Next, we will discuss intelligent agents that learn purposeful behavior from interactions with their environment. These agents are used for applications such as self-driving cars or interactive design of virtual and augmented realities. Indeed, in this course we relate synthetic Bayesian intelligent agents to natural intelligent agents such as the brain. You will be challenged to code Bayesian machine learning algorithms yourself and apply them to practical information processing problems.

News and Announcements

• (02-Dec-2024) Sometimes the online notebook viewer (NBviewer) for the lecture notes does not work. In that case, you can view the lecture notebooks straight at the github repository https://github.com/bertdv/BMLIP, since github has a built-in notebook viewer as well. In particular,

  • view the lecture notes here.
  • view the exercises here.

• (13-Nov-2024) Please sign up for Piazza (Q&A platform) at signup link. As much as possible we will use the Piazza site for new announcements as well.

Instructors

• Prof.dr.ir. Bert de Vries (email: bert.de.vries@tue.nl) is the responsible instructor for this course and teaches the lectures with label B.
• Dr. Wouter Kouw (w.m.kouw@tue.nl) teaches the probabilistic programming lectures with label W.
• Tim Nisslbeck, Sepideh Adamiat, Wouter Nuijten and Fons van der Plas are the teaching assistants.

Materials

In principle, you can download all needed materials from the links below.

Lecture Notes

The lecture notes are mandatory material for the exam:

• Bert de Vries, PDF bundle of all lecture notes for lessons B0 through B12, version 30-Oct-2024.
• Wouter Kouw, PDF bundle of all probabilistic programming lecture notes for lessons W1 through W4 version 30-Oct-2024.

Books

The following book is optional but very useful for additional reading:

• Christopher M. Bishop (2006), Pattern Recognition and Machine Learning.

Software

Please follow the software installation instructions. If you encounter any problems, please contact us in class or on Piazza.

Lecture notes, exercises, assignment and video recordings

You can access all lecture notes, assignments, videos and exercises online through the links below:

Exams & Assignments

Preparation

• Consult the Course Syllabus (lecture notes for 1st class) for advice on how to study the materials.

• When doing exercises from the above table, feel free to make use of Sam Roweis’ cheat sheets for Matrix identities and Gaussian identities.

• In addition to the materials in the above table, we provide two representative practice written exams:

  • 3-Feb-2022: exam ; answers ; calculations
  • 2-Feb-2023: exam ; answers ; calculations

Programming Assignments

• Programming assignments can be downloaded and submitted through the links in the above table.

Grading

• The final grade is composed of the results of assignments A1 (10%), A2 (10%), and a final written exam (80%). The grade will be rounded to the nearest integer.
• For the resit:
  • we use the same grading scheme: the resit written exam counts for 80%, and we will make available a new programming assignment for the resit that counts for 20%, see the table above.
  • If you don’t submit a resit programming assignment, we will use your results from this year’s assignments A1 and A2. Programming assignments from previous years do not count.
  • If you do submit your resit programming assignment, we will use your score for the resit assignment (and not your previous scores for A1 and A2 if you submitted those as well).