campbell@aero.org (Andrew Campbell) (04/11/91)
Los Angeles (LA) Association for Computing Machinery (ACM)
Technical Activity Committee on Artificial Intelligence (TACART)
TACART
TOPIC: Correcting Rule-Base Errors from Experience
SPEAKER: Prof. Michael Pazzani, UCI ICS Dept.
DATE: Wednesday, 24 April, 1991.
Note that this is the FOURTH Wednesday of the month.
TIME: The talk will begin promptly at 7:30 p.m.
There may be a few minutes of TACART business beforehand.
ABSTRACT: Techniques proposed for combining inductive and
explanation-based learning methods can also be used to
detect errors in rule-based expert systems, to isolate the
blame to a small number of rules, and to suggest revised rules.
FOCL - an extension of Quinlan's FOIL program - is a learning
system that can take advantage of domain knowledge (such as a
knowledge base) to learn classification rules, even if the
domain knowledge is incomplete and incorrect. A prototype
knowledge acquisition tool, KR-FOCL, can use FOCL traces to
suggest revisions to the knowledge base.
BIOGRAPHY: Michael Pazzani is an Assistant Professor of Information and
Computer Science at the University of California, Irvine.
He is the author of the book "Learning Causal Relationships"
and the accompanying machine learning software (OCCAM-lite)
published by Lawrence Erlbaum Associates, Hillsdale, NJ.
He received his PhD in 1988 from UCLA and is interested in
human and machine learning.
PLACE: The Conference Room, in the Basement of the Santa Monica
Public Library, located at the corner of 6th St. and
Santa Monica, in Santa Monica (4th/5th St. Exit from I-10).
DINNER: No dinner is planned. No social activities are planned.
CONTACT: L. Andrew Campbell (213) 336-8642 (days)
--
zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz
|L. Andrew Campbell internet: campbell@aerospace.aero.org |
|M1-102 PO Box 9295 organization: The Aerospace Corp. |
|Los Angeles CA 9009 telephone: (213) 336-8642 |