Proceedings

Flexible Learning of Problem Solving Heuristics Through Adaptive Search

Download BibTeX Cite

Proceedings of the Eighth International Joint Conference on Artificial Intelligence. Learning and Knowledge Acquisition. Pages 422-424.

© 1983 International Joint Conferences on Artificial Intelligence

Abstract

Noting th at th e methods employed by ex istin g learnin g sys tems are often bound to th r: intended task domain and have little applicabi lity outside that do main, tli is paper co nsiders an :.\lterna ti ve learning sys tem design th at offers g reater fl ex ibility with o ut sacrifi c ing performance. An operational prototype, co nstru cted around a powerful adap tive searc h technique, is presented and appli ed to th e problem of acquiring problem so lving heuristics through experi ence. S om e performance results obt ai ned with th e system in a poker betting do main are reported and compared with th ose of a pr ev iously in vesti ga ted learning sys tem in th e same domain . It is seen th at comp arable levels of performance are achieved by th e two system s. despite the latter 's dependence on a considerable a mount of domain specific knowl edge for effective operation.

  1. uchanon , B. and T.M Mitchell , " Moclei•Direc ted Learning of Prod uc tion Rul es", in Pattern -Oirec ted Inference Systems, Waterman and Hayes-Roth , Eel ., Academic Press, 1978.
  2. Carbonell , J.G ., " Experi enlial Learning in An alogical Problem Solving" , Proc. AAA /-82, Au g ust. 1982.
  3. DeJong,K .A ., "Adapti ve System Design: A Genetic Approach" , IEEE Trans . on Ma n, Sys tems, an d Cyberne tics 10, 9 (September, 1980) .
  4. DeJo nD , K.A. and T. Smith , !'Geneti c Al gorithms Appli ed to Inform ati on Dri ven Models of US Mi g ration Pattern s " , • 12th Annu al Pit tsbu rgh Co nt. on Modeling and Simulation , April , 1981 .
  5. Fogel, L.J ., A.J . Owe ns and M.J. Walsh, 1\rtificia ll ntellig ence Th rough Sim ula ted Evolution, Wiley, New York, 1966.
  6. Hed ri ck , C. L., " Learnin g Prod uction Sys tems from Exa mples" . Art ificial Intell ige nce 7 (1976).
  7. S. Smith 425
  8. Holland , J.H ., Aclap tatio n in Natural ancl Artific ial System s, Uni v. of Mic higan Press, 1975.
  9. Holl and , J.H. and J. Reitman, "Cognitive Systems Based on Adapti ve Algo rithms " , in Pa ttern-Direc ted In fe rence Systems, Waterm an ancl Hayes -Roth, Ed ., Academi c Press, 1Q78.
  10. Holl and . J.H., "Ad aptive Algorithms for Discovering and Using General Pattern s in Growing Knowledye Bases" , Int.
  11. Jo urn al on Po lic y Analysis and Informatio n Systems 4, 2 (1980).
  12. Mitchell , T.M., "Generalization as Search" , Artific ial Intellige nce 18 (1 982) .
  13. Mitchell , T.M. , P.E. Utgoff and R.B. Banerji, "Learning Problem Solving Heuristics by Experimentation" , in Ma c hine Learning, Michalski , Carbonell , and Mitchell , Ed ., Tioga Press, 1983.
  14. Smith , S. F., A Learning System Based on Genetic Adaptive Mgorithms, Ph.D . Th ., Univ . of Pittsburg h, December, 1980.
  15. Stolf o, S. J. and M.C. Harrison , " Automati c Discovery of Heu•dics for Nondetermini sti c Programs", Proc. /JCA I-79, 1979.
  16. Waterm an, O.A., "Generalization Learnin g Techniques for Aut u~ol ati n g th e Learn ing of Heuristics" , 1\ rtificia l Intellige nce 1 (1 970).

References were extracted automatically from the PDF and may contain errors.

Welcome Future Conferences AI Journal Division IJCAI Proceedings About IJCAI IJCAI Awards