Predoctoral Trainee

Kendrick Boyd Computer Sciences Graduate Program Advisor: David Page Email: boyd@cs.wisc.edu 262-6600

Statistical Relational Learning is proving to be a useful method of applying machine learning techniques to structured data residing in multiple tables such as that involved in medical diagnosis and prediction from clinical data. With the increasing availability of genetic information, methods must be found to integrate both the clinical and genetic data into the models. Davis et al (2008) used Marshfield clinical data and a learning technique called SAYU that lends itself to multi-table data to predict adverse drug
reactions. SAYU uses Alelph, an inductive logic programming (ILP) system to find potential clauses and then checks if those clauses should be added as variables to a Tree Augmented Naïve Bayes (TAN) model.
 
I will investigate whether more expressive Bayesian network structures and alternative rule searches during ILP can improve model accuracy for the SAYU model.  An extension of TAN would allow a much wider variety of Bayesian networks to be considered, potentially improving the models generated. On the ILP side, I will investigate using other methods of generating clauses than Aleph.

CV

2009 CV (.pdf format)

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.