NEW COURSE - SPRING 1998
"INTRODUCTION TO NEURAL COMPUTATION"
PHYX D60-0: SPECIAL TOPICS IN STATISTICAL PHYSICS
Time: MW 2:00-3:30 Room: 1384 Tech Instructor: Sara A. Solla
Office Address: 3308 Tech, Evanston Campus
Office Hours: M 3:30-5:00 or by appointment
FIRST MEETING: Monday, March 30 at 2:00PM in Tech 1384
COURSE DESCRIPTION: An introduction to "Neural Computation" aimed at first and second year graduate students and seniors in physics, applied and pure mathematics, computer science, electrical engineering, biomedical engineering, mechanical engineering, neuroscience, psychology, and cognitive sciences. The course will be based on neural network models, to be developed and analyzed and used as prototypes to investigate issues of learning and adaptation, including applications to brain modeling and pattern recognition. Topics to be discussed include: model neurons, neural networks, recurrent networks for associative memory, capacity and retrieval, multilayer perceptrons, learning algorithms, emergence of generalization ability, compact network architectures, generative models, and learning of time sequences.
PREREQUISITES: knowledge of linear algebra and some calculus
TEACHING METHOD: two lectures per week, 1.5 hours each
EVALUATION METHOD: homework problems and a take-home final examination
Introduction to the Theory of Neural Computation, by J. Hertz, A. Krogh, and R.G. Palmer.
Modeling Brain Function, by D.J. Amit;
An Introduction to Neural Networks, by J.A. Anderson;
Neural Networks: A Comprehensive Foundation, by S. Haykin;
Neural Networks for Pattern Recognition, by C.M. Bishop.