EPROTOS exemplar-based learning program

EPROTOS is an extension to Bareiss' PROTOS. (Actually, EPROTOS is an extension to CL-PROTOS which is the Lisp implementation of PROTOS by Dan Dvorak at UT Austin). The extension focused on enabling PROTOS to operate on cases described by a mix of discrete and numeric attributes. The extension employs ECOBWEB as a sub-module for learning and performing the "matching" on numeric attributes.

PROTOS itself is an exemplar-based learning systems that has some interesting mechanisms whose benefits have not been fully explored or appreciated.

EPROTOS was developed as part of Bridger--- an experimental learning system developed to explore the possibility of using machine learning techniques for the acquisition and improvement of design knowledge. The role of EPROTOS was the acquisition of redesign knowledge. Given a description of a cable-stayed bride that violates the design code, the system was supposed to retrieve a ranked list of the most appropriate redesign modification. Not too much testing was performed on this module but the initial results were encouraging.

If you want the CL-Protos code, check out Bruce Porter's ftp area at UT Austin.


A Sample Publications Include:


This page was accessed times since Oct. 1996. Current loading time
If you have any comments or questions, please send email to yoram@eng.tau.ac.il
Yoram Reich
Last modified: Mon Aug 11 18:42:41 IDT 1997