- Turing award (Nobel prize in Computer Science) for Bayesian networks (Judea Pearl)!
- Final marks exam (with seminar results); preliminary marks resit 2017
- Start course on Wednesday, 15th February, 2017; time 15:45-17:30, room 174
- Book used in the course: Bayesian Artificial Intelligence
- Assessment: exam (40%); assignment 1 and 2 (15% each); essay and seminar (30%) or, if you don't take part in the essay and seminar: exam (60%); assignment 1 and 2 (20% each)
- Peter Lucas is available for questions in office 123 on Wednesdays
- Practical assignment I and II are now online
- Deadline for assignment I: 30th April, 2017; for assigment II: 24th May, 2017
- Resit: Wednesday, 23rd August, 2017 (15:30-17:30, room 413) [mock exam] (solutions)
- Joint seminar meeting 28th June, 15:00-17:00, room 403
- No lecture on: 10th May, 2017 (just work on the preparation for the exam and assignment II)
-
Introduction: Reasoning under uncertainty and Bayesian networks (15th February, 2017)
[Slides PDF]
-
Bayesian networks: principles and definitions (22nd February, 2017)
[Slides PDF]
Reading: Chapters 1 and 2 from the "Bayesian Artificial Intelligence" book -
Building Bayesian networks (8th March, 2017)
[Slides: PDF]
Reading: Chapters 2 and 9 from the "Bayesian Artificial Intelligence" book -
Markov independence I & II (15th and 22nd March, 2017)
[Slides: PDF]
Background: Markov equivalence in Bayesian networks
Try running the Baseball algorithm: Netlogo software, Bayesball program -
Learning Bayesian networks (5th April, 2017)
[Slides: PDF]
Optional: Bayesian parameter learning -
Pearl's algorithms for probabilistic inference (12th April, 2017)
[Slides: PDF]
with further explanation
Reading: Chapter 3 from the "Bayesian Artificial Intelligence" book -
Other algorithms for probabilistic inference (19th April, 2017)
[Slides: PDF]
with further explanation
- See topics
-
Assignment I: build a Bayesian network in pairs; example networks and software (Deadline: 30th April, 2017)
- Assignment II: learning Bayesian networks (Deadline: 24th May, 2017):
- Database 1: iris dataset
- Database 2: data of patients with non-Hodgkin lymphoma (NA is missing value) and a mapping between dataset and NHL network download
For the manual NHL network see: example networks and software - Database 3: data of patients with breast cancer
For the manual BC network see: example networks and software
Bayesian Networks
|
|
Bayesian Networks is about
the use of probabilistic models (in particular Bayesian networks)
and related formalisms such as decision networks in problem solving, making
decisions, and learning.
Preliminary Schedule
Content of Lectures:
Seminar:
Content of Practicals:
Exercises:
Last updated: 13th January, 2017
Peter Lucas |
LIACS Staff
LIACS
Email: plucas AT liacs.nl
