BE 630 Elements of Neural Computation, Complexity and Learning

Prerequisite(s): A semester course in probability or equivalent exposure to probability (e.g. ESE 530). Non-linear elements and networks: linear and polynomial threshold elements, sigmoidal units, radial basis functions. Finite (Boolean) problems: acyclic networks; Fourier analysis and efficient computation; projection pursuit; low frequency functions. Network capacity: Feedforward networks; Vapnik-Chervnenkis dimension. Learning theory: Valiant's learning model; the sample complexity of learning. Learning algorithms: Perception training, gradient descent algorithms, stochastic approximation. Learning complexity: the intractability of learning; model selection.

Department of Bioengineering
School of Engineering and Applied Science
University of Pennsylvania
210 S. 33rd Street
Room 240 Skirkanich Hall
Philadelphia, PA 19104
Phone No.: (215) 898-8501
Fax No.: (215) 573-2071
beoffice@seas.upenn.edu

Send Comments and Suggestions to:
beweb@seas.upenn.edu