CAP 6615 Neural Networks (Spring 2011 )

Syllabus

Semester Credit Hours:

3

Number of Lecture Hours:

39 (50 minute) lectures

Examinations will be administered during three of our 42 class meetings hours.

Class Meeting Time and Room

MWF 3 (9:35 AM - 10:25 AM)
NPB 1200 (Physics Building)

Instructor

Joseph N. Wilson (jnw@cise.ufl.edu)
Room E472 CSE Bldg.
Phone: (352) 392-1360
Office Hours:

• 7th Period (1:55PM - 2:45PM) MW
  
• 4th Period (10:40AM - 11:35AM) T

Prerequisites:

CAP 5635 (Artificial Intelligence Concepts)

Good linear algebra and statistics background.

Textbook

Neural Networks and Learning Machines, 3/E
Simon Haykin, McMaster University, Ontario Canada

ISBN-10: 0131471392
ISBN-13: 9780131471399

Publisher: Prentice Hall
Copyright: 2009
Format: Cloth; 936 pp
Published: 11/18/2008

Course Objectives

Introduce fundamental concepts of neural networks and study several network models in detail. After taking this course, the student will be ready to understand the structure, design, and training of various types of neural networks and will be ready to apply them to the solution of problems in a variety of domains.

Course Description

The course will introduce students to the fundamental concepts behind neural networks: the biological motivation for their design, the practical developments that led to their evolution over time, and the mathematical basis for their applicability to problem solving domains. We start by looking at the single layer perceptron. We compare it's inherent problem solving capabilities to several other techniques. Then we consider the multilayer perceptron in detail. We follow that with a presentation of radial-basis function networks. We then consider the popular support vector machine model. Then we look at self-organizing maps. If time permits, we will consider the use of information theory in learning.

Course Requirements:

• Homework (10%)
• Computing projects (20%)
• Three Examinations (20% Each)
• Attendance & class participation (10%)

Course Outline by Topical Areas:

• Overview: neural networks, human brain, learning processes.
• Single-layer perceptron model
• Linear regrssion, maximum a-posteriori (MAP) estimation, and relation to neural networks
• Least-Mean-Square algorithm
• Multilayer perceptrons
• Radial-basis function networks
• PCA and Kernel PCA
• Self-organizing maps
• Information-theoretic learning models
• Clustering Methods