CHAIR OF

COMPUTATIONAL

INTELLIGENCE

General Information

This page contains information about the online lecture "Bayesian Networks". The course is held in winter term 2021/2022 by Prof. Dr. Rudolf Kruse and Dr. Alex Dockhorn.

With the increasing number of covid cases we have moved the Bayesian Network lecture as well as the exercise class from December to an online format.

Lecture Videos, Lecture slides, and  Exercise class tasks are available on our CICloud.

The Zoom Session starts every Wednesday at 1pm and allows for a discussion of the lecture's material as well as the exercise class tasks and solutions.

The written exam will take on 2.2.2022, 15-17 h, in room HS1

Announcements

You should have background knowledge on fundamentals of mathematics and computer science such as algorithms, data structures etc. Also, insights into probability theory are highly recommended.

Topics

• Representation of uncertain information
• Bayesian Networks
• Decomposition of spaces and separation of graphs
• Evidence propagation in probabilistic networks
• Learning of probabilistic networks
• Revision of probabilistic networks
• Decision Graphs and Influence Diagrams
• Causal Networks

 Note that Bayesian Networks is a master course, both lecture and exercise will be given in English.

Schedule

Lectures and Tutorials

Lectures: Prof. Dr. Rudolf Kruse, rudolf.kruse@ovgu.de

Tutorials: Dr.-Ing. Alexander Dockhorn,  alexander.dockhorn@ovgu.de

You find slides of the lectures below (they will be updated once a week). The slides and the lecture videos are available on our cicloud. 

All information about the tutorials (such as exercise sheets) are also  provided on our cloud  (see cicloud.cs.ovgu.de)

Conditions for  Exams

A new assignment sheet containing written and programming assignments is published every week. The written signments must be submitted.

Submitting a solution means that you are willing and able to explain and present the assignment and your solution proposal (which does not need to be completely correct)

during the online tutorial. However, you should be prepared thoroughly to present your solution.

The graded certificate for this course is issued to students who

• regularly contribute well in the exercises,
• submitted at least 2/3 of all written assignments
• presented at least twice a solution to a written assignment during the exercise (this number is reduced in case not everybody can present twice due to number of exercises)
  
• finally pass the exam after the course

Note that grade exams have to be officially announced to the examination office.

Slides

Note that the slides may be subject to change during the course, you'll find here the latest versions after the lectures. Recordings of last year will be made available on the CI Cloud. In case you forgot the password, please send an e-mail to alexander.dockhorn@ovgu.de.

Exercises

Materials for our Exercise Classes will be published on the CI Cloud. In case you forgot the password, please send an e-mail to alexander.dockhorn@ovgu.de

Additional Material

Feel free to check out the following supplementary material that augment the lecture and exercise.

• Illustration of Joint Tree Propagation
• Example Network used in the clique propagation lesson.
• Blood group determination of Danish Jersey cattle in the F-blood group system

Software

We recommend to use the HUGIN software for Bayesian Network and Inference Diagram technology. 

HUGIN LITE a free version for download.

LiteratureComputational Intelligence - A Methodological Introduction

   Kruse, R., Borgelt, C., Braune, C., Mostaghim, S., Steinbrecher, M., Computational Intelligence - A methodological introduction, 2. Edition, Springer-Verlag, London 2016

   This book is available via free download in the library. The 3. Edition will be available at Springer from February 2022