This is supplemental course information, designed to give you a fuller picture of the course and an expanded look at the topics covered. This is an unofficial document. The USC Course Catalog is the binding description of all university courses. Information such as books, materials covered, and the order of topics is subject to change. Please consult instructor for this semseter to get more upto date course information.
Summary:
The course presents modern neural networks and fuzzy systems. Lectures focus on the formal structure of fuzzy and neural systems. Project required that applies these techniques to signal processing, financial engineering, or other approved area.
Textbooks:
Required:
Kosko, B., Fuzzy Engineering, Prentice Hall, 1997
Kosko, B., Neural Networks and Fuzzy Systems, Prentice Hall, 1992
Note: Above two texts are available as a bound copy in the bookstore
Kosko, B., Heaven in a Chip, Random House, 2000.
Hagstrom, R. G., The Warren Buffett Way, second edition, Wiley, 1997
Recommended:
Fabozzi, F. J., Investment Management, ed., Prentice Hall, 1999
COURSE OUTLINE
Introduction to neural networks and fuzzy systems.
BAM stability. Multivalued truth and logic.
Neural signals. Set structure.
Subsethood. Partial equivalence.
MIDTERM I. Rational asset pricing.
Two-stage RAP analysis. Financial derivatives.
Standard additive model.
Unsupervised clustering. Supervised SAM learning.
Rule structure and system statistics. Project proposals due.
Gradient systems. General neural stability.
MIDTERM II. Feedback SAMs.
LMS. Training multilayer perceptrons.
Theories of learning and reinforcement.
Project presentations.
Project presentations.
Professor Kosko email: kosko@sipi.usc.edu
