The FERMI System: Inducing Iterative Macro-Operators from Experience

Patricia W. Cheng, Jaime G. Carbonell

Automated methods of exploiting past experience to reduce search vary from analogical transfer to chunking control knowledge. In the latter category, various forms of composing problem-solving operators into larger units have been explored. However, the automated formulation of effective macro-operators requires more than the storage and parametrization of individual linear operator sequences. This paper addresses the issue of acquiring conditional and iterative operators, presenting a concrete example implemented in the FERMI problem-solving system. In essence, the process combines empirical recognition of cyclic patterns in the problem-solving trace with analytic validation and subsequent formulation of general iterative rules. Such rules can prove extremely effective in reducing search beyond linear macro-operators produced by past techniques.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.