At the end of this course, the student should be able to:
- Goal 1. Demonstrate knowledge of the theoretical foundations of probabilistic models:
- understand and apply the basic principles of probability theory;
- being able to understand and use properties of various probabilistic graphical models.
- Goal 2. Develop problem-solving skills applicable to a range of domains:
- understand the principles of building Bayesian networks from expert knowledge and from data, and apply them with the aid of software tools;
- use various (exact and approximate) inference algorithms for probabilistic reasoning;
- being able to identify methods that are appropriate for building and evaluating realistic Bayesian and decision models.
- Goal 3. Demonstrate competence and skills necessary to conduct successful scientific investigations:
- critically analyse a specialised topic from the field of probabilistic models on the basis of relevant scientific literature;
- work effectively in a team with students to discuss and argue on a scientific topic;
- Effectively and clearly communicate, both orally and in writing, research insights and ideas to scientific audiences.
Bayesian networks are powerful, yet intuitive tools for knowledge representation, reasoning under uncertainty, inference, prediction, and classification. The history of Bayesian Networks dates back to the groundbreaking work of Judea Pearl and others in the late 1980s, for which Pearl was given the Turing Award in 2012. Since then, Bayesian networks have evolved to become key parts of the data scientist's toolbox and are used in many application domains, notably medicine and molecular biology. This course will cover the necessary theory to understand, build, and work with Bayesian networks. Practical work will focus on implementing Bayesian networks in real-world domains.
|There is no compulsory book for this course, the lecture notes will be self-contained. However, it is of course recommended to consult other sources during self-study, such as the books listed below. The first two books are available as e-books through the university library.|
* R.G. Cowell, A.P. Dawid, S.L. Lauritzen and D.J.Spiegelhalter, Probabilistic Networks and Expert Systems, Springer, New York, 1999.
* F.V. Jensen and T. Nielsen, Bayesian Networks and Decision Graphs, Springer, New York, 2007
* K.B. Korb and A.E. Nicholson, Bayesian Artificial Intelligence, Chapman & Hall, Boca Raton, 2004 or 2010
* C.M. Bishop, Pattern Recognition and Machine Learning, Springer, 2006
* J. Pearl, Causality (2nd edition), Cambridge University Press, 2009
• 6 hours guided group project work
• 24 hours lecture
• 40 hours group project work without guidance
• 2 hours student presentation
• 96 hours individual study period
|The course is part of the Computing Science Data Science theme and also part of the AI master programme.|
|* General Introduction|
* Conditional Independence
* Managing Network Complexity
* Structural Equation Models
* Latent Variables
* Exact Inference
* Approximate Inference
* Markov Equivalence
* Structure Learning
* Non-DAG Models
* Causality and Interventions
|Written exam, two project reports & code, a presentation.|
|Previous knowledge in probability and statistics may help to digest the content more quickly, but the necessary bits such as the basic laws of probability and conditional probability will also be recapped at the beginning of the course.|
|Johannes Textor, ICiS / Department of Tumor Immunology, email@example.com|