Gregory Wheeler
Department of Computer Science
Universidade Nova de Lisboa
grw@fct.unl.pt

 

course description:

Probability and logic are two different formalisms that are suited to representing complementary types of information. Various types of uncertainty can be captured by a probability distribution, or sets of distributions, whereas logic affords one the ability to identify types and relations, and to exploit information lodged in the structure of sentences.

Unfortunately, it is difficult to combine these formalisms for the purposes of making informative, tractable inference. This course surveys various types of semantics on offer for probabilistic logics, and considers the use of probabilistic networks for effecting tractable inference.

course materials:

• Rolf Haenni, Jan-Willem Romeijn, Gregory Wheeler, Jon Williamson. "Probabilistic Logics and Probabilistic Networks", to appear.
  

web link: PROGICNET.

assessment

Grades are determined by

• 4 analytical exercises (40% +/- 3%)
• project (20% +/- 3%)
• a final exam (40% +/- 3%).

No late assignments are accepted.