An Analytic Learning System for Specializing Heuristics

This paper describes how meta-level theories are used for analytic learning in M U L T I - T A C . M U L T I - T A C operationalizes generic heuristics for constraint-satisfaction problems, in order to create programs that are tailored to specific problems. For each of its generic heuristics, M U L T I - T A C has a meta-theory specifically designed for operationalising that heuristic. We present examples of the specialisation process and discuss how the theories influence the tractability of the learning process. We also describe an empirical study showing that the specialised programs produced by M U L T I - T A C compare favorably to hand-coded programs.

• Type: Conference
• Year: 1993
• Where: IJCAI
• Authors: Steven Minton
• Keywords: Analytic Learning, Generic Heuristics, IJCAI 1993, IJCAI 2007, Meta-level Theories