EPIA'03 - 11th Portuguese Conference on Artificial Intelligence

	Session: December 7, 11:30-12:15, Room B
Title:	On the Implementation of an ILP System with Prolog
	Nuno Fonseca, Fernando Silva, Rui Camacho, Vitor Santos Costa
Abstract:	Inductive Logic Programming systems (ILP) is a Machine Learning technique that has been quite successful in knowledge discovery for relational domains. ILP systems implemented in Prolog challenge the limits of Prolog systems due to heavy usage of resources such as database accesses and memory usage, and to very long execution times. In this work we discuss the fundamental performance issues found in an ILP engine -- the April system. Namely, we evaluate the impact of a fundamental technique, called coverage caching, that stores previous results in order to avoid recomputation. To understand the results obtained we profiled April's execution and present initial results. We advocate that the indexing mechanisms used in YAP Prolog database are inefficient and that improvement of these mechanisms may lead to significant improvements in Prolog based ILP systems.

Back to schedule.