690 Machine Self Organization
and  Learning

            

Course Description

This course discusses advanced concepts of self-organizing networks of sparsely connected processing components (neurons). Neural-net implementations of pattern recognition algorithms provide important, practical advantages by allowing fast realization of parallel, iterative procedures. Operations of self-organizing neural networks will be developed and used for different neural functions. These functions will include implementation of associative memories, statistical self-organization and learning, and self-organization for the reinforcement lerning. An example self-organizing neural system simulating biological systems will be examined. The majority of biological intelligence creatures are simple, yet they can achieve complex information processing and computational tasks that current artificial intelligence cannot match.  Can we use these simple models to learn how to design better artificial intelligence? 

The emphasis in this course is on development of the concept of self-organizing neural system with locally interconnected processing components.  Students will simulate neural networks for patter recognition and classification using PC software tools. This course will prepare you to study computational principles and hardware organization of what we mean by intelligence and goal-directed behavior.  How to motivate machine to act on its own, yet to satisfy a desired objective?  How machine interaction with environment leads to better behavior, better understanding, and success in its mission? What are the computational and hardware issues in doing this efficiently and in realtime? We try to define what it means to be intelligent, anticipate, learn from experience, make associations,  perceive, act independently, self evaluate and think.

Syllabus
Schedule
Resources
Grades
Reference Books

Related links

Reinforcement Learning Repository University of Massachusetts, Amherst
USC Brain Theory and Artificial Intelligence CS 564 : Fall 2001
Bio-inspired Computing AI 23
Introduction to Reinforcement learning
Neural Nets by Kevin Gurney
Michael Arbib Homepage