Abstract
Soar is an architecture for a system that is intended to be capable of general intelligence. Chunking, a simple experience-based learning mechanism, is Soar's only learning mechanism. Chunking creates new items of information, called chunks, based on the results of problem-solving and stores them in the knowledge base. These chunks are accessed and used in appropriate later situations to avoid the problem-solving required to determine them. It is already well-established that chunking improves performance in Soar when viewed in terms of the subproblems required and the number of steps within a subproblem. However, despite the reduction in number of steps, sometimes there may be a severe degradation in the total run time. This problem arises due to expensive chunks, i.e., chunks that require a large amount of effort in accessing them from the knowledge base. They pose a major problem for Soar, since in their presence, no guarantees can be given about Soar's performance.
In this article, we establish that expensive chunks exist and analyze their causes. We use this analysis to propose a solution for expensive chunks. The solution is based on the notion of restricting the expressiveness of the representational language to guarantee that the chunks formed will require only a limited amount of accessing effort. We analyze the tradeoffs involved in restricting expressiveness and present some empirical evidence to support our analysis.
Article PDF
Similar content being viewed by others
References
Brownston, L., Farrell, R., Kant, E. and Martin, N. (1985). Programming expert systems in OPS5: An introduc-tion to rule-based programming. Reading, MA: Addison-Wesley.
Chalasani, P. and Altmann, E. (1989). Comparing the representations in Soar and Theo. Unpublished, School of Computer Science, Carnegie Mellon University.
Chase, M.P., Zweben, M., Piazza, R.L., Burger, J.D., Maglio P.P. and Hirsh, H. (1989). Approximating learned search control knowledge. Proceedings of International Workshop on Machine Learning (pp.218-220).
Cohen, W. (1988). Generalizing number and learning from multiple examples in explanation-based learning. Proceedings of the Fifth International Conference on Machine Learning (pp.256-269).
Cohen, W., Mostow, J. and Borgida, A. (1988). Generalizing number in explanation-based learning. Proceedings of the Spring Symposium on Explanation-Based Learning (pp.68-72).
DeJong, G.F. and Mooney, R. (1986). Explanation-based learning: An alternative view. Machine Learning, I, 145-176.
Forgy, C.L. (1982). Rete: A fast algorithm for the many pattern/many object pattern match problem. Artificial Intelligence, 19, 17-37.
Forgy, C.L. (1984). The OPS83 Report (Technical Report CMU-CS-84-133) Pittsburgh, PA: Carnegie Mellon University, Computer Science Department.
Garey, M.R. and Johnson, D.S. (1978). Computers and intractability: A guide to the theory of NP-completeness. San Francisco, CA: W.H. Freeman and Company.
Greiner, R. and Likuski, J. (1989). Incorporating redundant learned rules: A preliminary formal analysis of EBL. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (pp.744-749).
Gupta, A. (1986). Parallelism in production systems. Doctoral dissertation, Computer Science Department, Carnegie Mellon University. Also a book, Morgan Kaufmann,(1987).
Gupta, A. and Tambe, M. (1988). Suitability of message passing computers for implementing production systems. Proceedings of the Seventh Conference on Artificial Intelligence (pp.687-692).
Hsu, W., Prietula, M. and Steier, D. (1989). Merl-Soar: Scheduling within a general architecture for intelligence. Proceedings of the Third International Conference on Expert Systems and the Leading Edge in Production and Operations Management (pp.467-481).
Iba, G.A. (1989). A heuristic approach to the discovery of macro-operators. Machine Learning, 3, 285-318.
Ishida, T. (1988). Optimizing rules in production system programs. Proceedings of the Seventh National Conference on Artificial Intelligence (pp.699-704).
Keller, R. (1987). Defining operationality for explanation-based learning. Proceedings of the Sixth National Conference on Artificial Intelligence (pp.482-487).
Laird, J.E., Swedlow, K.R., Altmann, E.M., Congdon, C.B. and Wiesmeyer, M. (1989). Soar 4.5 user's manual. School of Computer Science, Carnegie Mellon University and Department of Electrical Engineering and Computer Science, University of Michigan, June,1989.
Laird, J.E., Newell, A. and Rosenbloom, P.S. (1987). Soar: An architecture for general intelligence. Artificial Intelligence, 33 1-64.
Laird, I.E., Rosenbloom, P.S. and Newell, A. (1986). Chunking in Soar: The anatomy of a general learning mechanism: Machine Learning, 1, 11-46.
Levesque, H.J. and Brachman, R.J. (1985). A fundamental tradeoff in knowledge representation and reasoning. In Brachman, R.J. and Levesque, H.J. (Eds.), Readings in Knowledge Representation. Los Altos, CA: Morgan Kaufmann Publishers, Inc.
Markovitch, S. and Scott, P.D. (1988). The role of forgetting in learning. Proceedings of the Fifth International Conference on Machine Learning (pp.459-465).
Markovitch, S. and Scott, P.D. (1989). Information filters and their implementation in the SYLLOG system. Proceedings of the Sixth International Workshop on Machine Learning (pp.404-407).
Markovitch, S. and Scott, P.D. (1989). Utilization filtering: a method for reducing the inherent harmfulness of deductively learned knowledge. Proceedings of the Eleventh International Joint Conference on Artificial Intelli-gence (pp.738-743).
Minton, S. (1985). Selectively generalizing plans for problem-solving. Proceedings of the Ninth International Joint Conference on Artificial Intelligence (pp.596-599).
Minton, S. (1988). Quantitative results concerning the utility of explanation-based learning. Proceedings of the Seventh National Conference on Artificial Intelligence (pp.564-569).
Minton, S. (1988). Learning Effective Search Control Knowledge: An explanation-based approach. Doctoral dissertation, Computer Science Department, Carnegie Mellon University.
Minton, S., Carbonell, J.G., Knoblock, C.A., Kuokka, D.R., Etzioni, O. and Gil, Y. (1989). Explanation-based learning: A problem solving perspective. Machine Learning, Vol.40 (pp.63-118).
Miranker, D.P. (1987). Treat: A better match algorithm for AI production systems. Proceedings of the Sixth National Conference on Artificial Intelligence (pp.42-47).
Mitchell, T.M., Allen, J., Chalasani, P., Cheng, J., Etzioni, O., Ringuette, M. and Schlimmer, J.C. (1989). Theo: A framework for self-improving systems. In VanLehn, K. (Ed.), Architectures for Intelligence. Hillsdale, NJ: Lawrence Erlbaum Associates.
Mitchell, T.M., Keller, R.M. and Kedar-Cabelli, S.T. (1986). Explanation-based generalization: A unifying view. Machine Learning, I, 47-80.
Mooney, R. (1989). The effect of rule use on the utility of explanation-based learning. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (pp.725-730).
Nayak, P., Gupta, A. and Rosenbloom, P. (1988). Comparison of the Rete and Treat production matchers for Soar (A summary). Proceedings of the Seventh National Conference on Artificial Intelligence (pp.693-698).
Newell, A. (1990). Unified theories of cognition. Cambridge, MA: Harvard University Press. In press.
Newell, A. and Rosenbloom, P.S. (1981). Mechanisms of skill acquisition and the law of practice. In Anderson, J.R. (Ed.), Cognitive Skills and Their Acquisition. Hillsdale, NJ: Lawrence Erlbaum Associates.
Newell, A., Rosenbloom, P.S. and Laird, J.E. (1990). Symbolic architectures for cognition. In M.I. Posner (Ed.), Foundations of Cognitive Science. Cambridge, MA: Bradford Books/MIT Press. In press.
Oflazer, K. (1987). Partitioning in Parallel Processing of Production Systems. Doctoral dissertation, Computer Science Department, Carnegie Mellon University.
Pereira, L.M. and Porto, A. (1982). Selective backtracking. In Clark, K.L. and Tarnlund, S.A. (Eds.), Logic Programming. NY: Academic Press.
Polk, T.A. and Newell, A. (1988). Modeling human syllogistic reasoning in Soar. Proceedings of the Annual Conference of the Cognitive Science Society (pp.181-187).
Preiditis, A. and Mostow, J. (1987). PROLEARN: Towards a Prolog interpreter that learns. Proceedings of the Sixth National Conference on Artificial Intelligence (pp.494-498).
Reich, Y. (1988). Learning plans as a weak method for design. Department of Civil Engineering, Carnegie Mellon University, Unpublished.
Rosenbloom, P.S. and Laird, J.E. (1986). Mapping explanation-based generalization onto Soar. Proceedings of the Fifth National Conference on Artificial Intelligence (pp.561-567).
Rosenbloom, P.S. and Newell, A. (1986). The chunking of goal hierarchies: A generalized model of practice. In Laird, J.E., Rosenbloom, P.S., and Newell, A. (Eds.), Universal Subgoaling and Chunking: The Automatic Generation and Learning of Goal Hierarchies. Boston, MA: Kluwer Academic Publishers.
Rosenbloom, P.S., Laird, J.E., McDermott, J., Newell, A. and Orciuch, E. (1985). Rl-Soar: An experiment in knowledge-intensive programming in a problem-solving architecture. IEEE Transactions on Pattern Analysis and Machine Intelligence, 7, 561-569.
Scales, D.J. (1986). Efficient matching algorithms for the Soar/OpsS production system (Technical Report KSL-86-47). Palo Alto, CA: Stanford University, Knowledge Systems Laboratory, Department of Computer Science.
Shavlik, J. (1989). Acquiring recursive concepts with explanation-based learning. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (pp.688-693).
Shavlik, J.W. and DeJong, G.F. (1987). An explanation-based approach to generalizing number. Proceedings of the Tenth International Joint Conference on Artificial Intelligence (pp.236-238).
Shell, P. and Carbonell, J. (1989). Towards a general framework for composing disjunctive and iternative macro-operators. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (pp.596-602).
Smith, D.E. and Genesereth, M.R. (1986). Ordering conjunctive queries. Artificial Intelligence, 26, 171-215.
Steier, D.M. (1986). Speeding up Soarware. School of Compuer Science, Carnegie Mellon University, Unpublished.
Steier, D.M. (1987). Cypress-Soar: A case study in search and learning in algorithm design. Proceedings of the Tenth International Joint Conference on Artificial Intelligence (pp.327-330).
Steier, D.M., Laird, J.E., Newell, A., Rosenbloom, P. S, Flynn, R.A., Golding, A., Polk, T.A., Shivers, O.G., Unruh, A. and Yost, G.R. (1987). Varieties of learning in Soar: 1987. Proceedings of the Fourth International Workshop on Machine Learning (pp.300-311).
Tambe, M. (1988). Speculations on the computational effects of chunking. Computer Science Department, Carnegie Mellon University, Unpublished.
Tambe, M. and Newell, A. (1988). Some chunks are expensive. Proceedings of the Fifth International Conference on Machine Learning (pp.451-458).
Tambe, M. and Rosenbloom, P. (1988). Eliminating expensive chunks (Technical Report CMU-CS-88-189). Pittsburgh, PA: Carnegie Mellon University, Computer Science Department.
Tambe, M. and Rosenbloom, P. (1989). Eliminating expensive chunks by restricting expressiveness. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (pp.731-737).
Tambe, M. and Rosenbloom, P. (1990). A framework for investigating production system formulations with polynomially bounded match. Proceedings of the Eighth National Conference on Artificial Intelligence. (To appear.)
Tambe, M., Kalp, D., Gupta, A., Forgy, C.L., Milnes, B.C. and Newell, A. (1988). Soar/PSM-E: Investigating match parallelism in a learning production system. Proceedings of the ACM/SIGPLAN Symposium on Parallel Programming: Experience with applications, languages, and systems (pp.146-160).
Ullman, J.D. (1982). Principles of database systems. Rockville, MD: Computer Science Press.
Washington, R. and Rosenbloom, P.S. (1988). Applying problem solving and learning to diagnosis. Knowledge Systems Laboratory, Stanford University, December,1988, Unpublished.
Wiederhold, G. (1987). File organization for database design. NY: McGraw-Hill.
Yost, G.R. and Altmann, E.M. (1989). TAQL 3.0: Soar task acquisition system user's manual. School of Computer Science, Carnegie Mellon University, December,1989, Unpublished.
Yost, G.R. and Newell, A. (1989). A problem space approach to expert system specification. Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (pp.621-627).
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Tambe, M., Newell, A. & Rosenbloom, P.S. The Problem of Expensive Chunks and its Solution by Restricting Expressiveness. Machine Learning 5, 299–348 (1990). https://doi.org/10.1023/A:1022603523533
Issue Date:
DOI: https://doi.org/10.1023/A:1022603523533