Are Bayesian Networks Sensitive to Precision of their Parameters? (Expired)

Day: Wednesday April 13th
Time: 10:30am-11:30am
Room: ED 191

Speaker: Marek J. Druzdzel School of Information Sciences University of Pittsburgh
http://www.pitt.edu/~druzdzel

Title: Are Bayesian Networks Sensitive to Precision of their Parameters?

Abstract:
A popular belief has it that the accuracy of reasoning based on Bayesian networks is not too sensitive to precision of their parameters. This belief is based mainly on anecdotal evidence, supported by a limited number of empirical studies. I will present the results of a series of experiments focusing on this question in the context of a medical decision support system Hepar, based on a Bayesian network model. The results of these experiments are encouraging: Noise seems to have only a limited influence on accuracy. Precision of parameters, tested by their systematic roughening has also minimal influence on accuracy.


Breaking News


Recent News


Do You Have News for the Department of Computer Science?

  • Please send a plain text version of your posting, which can include URL links, to deptsec@cs.uregina.ca
RSS
To Top of Page