Bayesian networks and their applications
(lectures within the Postgraduate Programme 
at UNED Madrid in July 2007)

1. Inference with Bayesian networks
   (PDF of the presentation and LaTex source files):
   - basic tasks solved with Bayesian networks
     (computation of marginal probability, conditional probability, most probable configuration)
   - junction tree method
   - demo of Hugin (using example "Visit to Asia") 
   - excercise: creating a simple Bayesian network model using Hugin Lite and solving basic tasks using it
2. Classification 
   (PDF of the presenation and LaTex source files):
   - linear regression
   - logistic regression
   - naive Bayes classifier
   - Tree augmented Naive Bayes
   - demo of Weka System
   - excercise: learning a classifier for a given dataset using the Weka System
3. Learning Bayesian networks:
   (PDF of the presentation on score based learning, LaTex source files, and
   PDF of the presentation of EM-algorithm (slides of F.V. Jensen and T.D. Nielsen))
   - learning parameters of Bayesian Networks (EM-algorithm for incomplete data)
   - testing conditional independence
   - PC-algorithm
   - maximizing a criteria (e.g., BIC) 
   - equivalence classes of Bayesian networks, essential graphs
   - Greedy equivalence search (GES) algorithm
   - excercise: learning a Bayesian network model from a dataset using Hugin Lite
4. Decision-theoretic troubleshooting
   (PDF of the presentation and LaTex source files)
   - problem causes, solution actions, observations, their probabilities and costs
   - troublehooting strategy   
   - demo of Dezide troubleshooter
   - excercise: creating a simple troubleshooting model with Dezide Author
5. Computerized adaptive testing (CAT) using Bayesian networks
   (PDF of the presentation and LaTex source files)
   - Rasch model
   - adaptive versus fixed tests
   - student model and evidence models
   - optimal and myopically optimal tests
   - example of a model of students solving tasks with fractions  
   no exercise