I'm interested in artificial intelligence, especially machine learning, approximate reasoning, and constraint processing. I try to stay actively engaged in all three research areas.
First, I'm using various supervised and unsupervised machine learning methods, including symbolic, neural, and Bayesian approaches, to model the biological activity of chemical compounds in animals, humans, and the environment, including toxic and/or undesirable effects such as carcinogenicity and atmospheric ozone reactivity. A system we designed and implemented for predicting chemical carcinogenicity showed the best performance of six computer-based systems for predicting the carcinogenicity of a standard set of test chemicals.
Second, I've used new techniques in fuzzy logic and fuzzy constraint processing to develop improved intelligent multiagent process control systems. Our methods are not only more richly expressive than most other fuzzy control systems, but also employ useful new basic techniques for approximate reasoning.
work on intelligent support for concurrent engineering using
constraint processing. In this approach a constraint may be any expression in
first-order predicate calculus. This work has resulted in the development of a
language and runtime system with broad application to concurrent engineering
and other aspects of design.