CSCI 252 - Neural Networks and Graphical Models

Credits: 3
Planned Offering: Offered when interest is expressed and departmental resources permit.

Prerequisite: CSCI 112. A survey of the major developments in neural and belief networks, from the early perception models of the 1940s through the probabilistic Bayesian networks that are a “hot topic” in artificial intelligence today. Topics include the back-propagation algorithm, simple recurrent networks, Hopfield nets, Kohonen’s Self-Organizing Map, learning in Bayesian networks, and Dynamic Bayesian Networks, with readings from both popular textbooks and the scholarly literature. A major focus of the course is on writing programs to implement and apply these algorithms. Levy.