SPECIAL Bisc Seminar, 29 May, 4-5:00pm, 310 Soda Hall

Michael Lee (leem@cs.berkeley.edu)
Mon, 2 Jun 1997 18:04:31 +0200


This is a multi-part message in MIME format.

--------------65322D8A736B
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

-- 
_______________________________________________________________________
Michael A. Lee
Berkeley Initiative in Soft Computing
387 Soda Hall                                      Tel: +1-510-642-9827
Computer Science Division                          Fax: +1-510-642-5775
University of California                    Email: leem@cs.berkeley.edu
Berkeley, CA 94720-1776 USA       WWW: http://www.cs.berkeley.edu/~leem
_______________________________________________________________________

--------------65322D8A736B Content-Type: text/plain; charset=us-ascii; name="mitaim.5.29.97" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="mitaim.5.29.97"

Neural Fuzzy Agents for Profile Learning and Database Search

BISC Seminar

Sanya Mitaim Signal & Image Processing Institute Department of Electrical Engineering--Systems University of Southern California E-mail: mitaim@sipi.usc.edu

29 May 1997 310 Soda Hall 4:00-5:00pm

Abstract:

How can an intelligent agent learn what you like and dislike? The talk will show how neural fuzzy systems can approximate user profiles and can help search an image database. The user profile is an unknown preference map or bumpy utility surface defined over the space of search objects. Hilltops on the surface define well-liked search objects and valleys define well-disliked objects. The fuzzy agent system gives a finer and finer approximation of the user's profile as learning unfolds in a question-answer session and as the fuzzy rule patches move to cover the extrema or bumps of the sampled preference map. The fuzzy agent system uses a new adaptive equivalence measure to search for preferred objects. The search system adapts its associative search metric by moving set points closer to one another in fuzzy unit hypercubes of high dimension. The talk will discuss how such fuzzy agent systems can apply to image databases and how the authors arrived at the underlying mathematical tools. Formal aspects of the talk will appear in the upcoming special issue of Presence on Autonomous Agents, Adaptive Behaviors, and Distributed Simulations. This research is a joint effort with Professor Bart Kosko.

---------------------------------------------------------------------------- CS-Seminars@CS.Berkeley.EDU

--------------65322D8A736B--