Predoctoral Trainee

Eric Lantz Computer Sciences Doctoral Program Faculty Supervisor: David Page Email: lantz@cs.wisc.edu 262-6601

Probabilistic graphical networks such as Bayesian networks have been applied to a variety of tasks, including predicting gene interaction networks. Searching the complete space of possible networks is intractable for problems of interesting size, so simplifications must be made. However, these simplifications also restrict the types of relationships that can be learned exclusively from data. One type of relationship we would like to be able to elucidate is “correlation immune” functions, which are characterized by the inability to detect any pairwise correlation between relevant variables. We augment a commonly used Bayesian network structure learning algorithm called sparse candidate with a technique called skewing to randomly permute the data distribution multiple times, causing relevant variables to stand out against irrelevant variables. Good results have been achieved with synthetic data, and we will fit to appropriate biological datasets.

CV

Computer Sciences web page: http://www.cs.wisc.edu/~lantz

Publications

Search for publications by Eric Lantz(Pub Med)

Feedback, questions or accessibility issues.
Graduate Program  |  Postdoctorate Program  |  Short Term Research Program

File last modified: August 13, 2003
Copyright © 2003 The Board of Regents of the University of Wisconsin System.