Massively Parallel Computational Models of Biological Systems

Abstract Conventional computer programs have been used for some time as models of intelligent behavior, with mixed success. Everyone acknowledges that animal brains are different from electronic machines in speed, concurrency, reliability, and flexibility. The problem is to develop methodologies that allow one to simulate more realistic models of intelligent behavior. Our group has been active in computer vision for many years and has been working on massively parallel models for the last four years. The talk will cover a formalism for massively parallel models, a general simulator for such models and their applications to problems of vision and natural language parsing.