×

Multilevel counterfactuals for generalizations of relational concepts and productions. (English) Zbl 0445.68067

Summary: In the induction of relational concepts and productions from examples and counterexamples, a ‘counterfactual’ is a set of conditions which must be false if a generalization is to be satisfied. Multilevel counterfactuals may themselves contain counterfactuals nested to any level, providing a series-like concept representation mechanism. An algorithm is presented, with correctness proof, which computes multilevel counterfactuals by recursively reducing the original induction problem to a smaller ‘residual’ problem, whose generalization gives the desired counterfactual. Winston’s empirical method for determining ‘must-not’ conditions (single level counterfactuals) is shown to yield erroneous results in certain rather ordinary circumstances. Computed examples are presented for the generalization of complex geometric scenes and the learning of blocksworld operators without resorting to the common CLEARTOP expedient.

MSC:

68T05 Learning and adaptive systems in artificial intelligence
Full Text: DOI

References:

[1] Anderson, J. R., Induction of augmented transition networks, Cognitive Sci., 1, 125-158 (1977)
[2] Brayer, J. M.; Fu, K. S., Some multidimensional grammar inference methods, (Chen, C. H., Pattern Recognition and Artificial Intelligence (1976), Academic Press: Academic Press London), 29-60
[3] Bruner, J. S., (A Study of Thinking (1956), Wiley: Wiley New York)
[4] Hayes-Roth, F.; McDermott, J., An interference matching technique for inducing abstractions, Comm. ACM, 21, 5, 401-410 (May 1978) · Zbl 0372.68025
[5] Hedrick, C. L., Learning production systems from examples, Artificial Intelligence, 7, 21-49 (1976)
[6] Hunt, E. B., (Experiments in Induction (1966), Academic Press: Academic Press London)
[7] Larson, J.; Michalski, R. S., Inductive inference of VL decision rules, (Proceedings of the Workshop on Pattern-Directed Inference Systems. Proceedings of the Workshop on Pattern-Directed Inference Systems, SIGART Newsletter (June 1977)), 38-44
[8] Mitchell, T. M., Version spaces: A candidate elimination approach to rule learning, (Fifth Int. Joint Conf. on Artificial Intelligence. Fifth Int. Joint Conf. on Artificial Intelligence, Cambridge, MA (1977)), 305-310
[9] Newell, A.; Simon, H. A., (Human Problem Solving (1972), Prentice-Hall: Prentice-Hall Englewood Cliffs, NJ)
[10] Plotkin, G. D., Automatic methods of inductive inference, (Experimental Programming Reports No. 24 (1971), Dept. of Machine Intelligence and Perception, Univ. of Edinburgh) · Zbl 0219.68045
[11] Simmons, R. F., Rule-based computations on English, (Waterman, D. A.; Hayes-Roth, F., Pattern-Directed Inference Systems (1978), Academic Press: Academic Press London), 455-468
[12] Vere, S. A., Induction of concepts in the predicate calculus, (Fourth Int. Joint Conf. on Artificial Intelligence. Fourth Int. Joint Conf. on Artificial Intelligence, Tbilisi, USSR (1975)), 281-287
[13] Vere, S. A., Relational production systems, Artificial Intelligence, 8, 47-68 (1977) · Zbl 0347.68050
[14] Vere, S. A., Inductive learning of relational productions, (Waterman, D. A.; Hayes-Roth, F., Pattern-Directed Inference Systems (1978), Academic Press: Academic Press London), 281-295
[15] Vere, S. A., Induction of relational productions in the presence of background information, (Fifth Int. Joint Conf. on Artificial Intelligence. Fifth Int. Joint Conf. on Artificial Intelligence, Cambridge, MA (1977)), 349-355
[16] Winston, P. H., Learning structural descriptions for examples, MIT AI Lab. Tech. Rept. 231 (September 1970)
[17] Young, R. M., Analysis of an extended concept-learning task, (Fifth Int. Joint Conf. on Artificial Intelligence. Fifth Int. Joint Conf. on Artificial Intelligence, Cambridge, MA (1977)), 285
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.