Hypothesis-driven Constructive Induction: AQ17-HCI

Traditional concept learning methods express the learned hypothesis using descriptors that are present in describing the training examples. In other words, they learn in the same representation space in which training examples are presented. For many practical problems this is a serious limitation, because concepts to be learned require descriptors that go beyond those originally provided.

To attack such problems, a constructive induction approach splits the learning process into two intertwined searches: one-for the most appropriate representation space for the given learning problem, and second -for the best inductive hypothesis in the newly created space.

A hypothesis-driven constructive induction method changes the concept representation spaces in the process of the concept learning. The changes involve expansion and contraction of the representation space, and are based on the analysis of consecutively created inductive hypotheses.

Arciszewski, T., Michalski, R.S., Wnek, J., "Constructive Induction: the Key to Design Creativity," Proceedings of the Third International Round-Table Conference on Computational Models of Creative Design, Heron Island, Queensland, Australia, December 3-7, 1995.

Szczepanik, W., Arciszewski, T. and Wnek, J., "Empirical Performance Comparison of Two Symbolic Learning Systems Based On Selective And Constructive Induction," Proceedings of the IJCAI- 95 Workshop on Machine Learning in Engineering, Montreal, Canada, August, 1995.

Michalski, R.S. and Wnek, J., "Learning Hybrid Descriptions," Proceedings of the 4th International Symposium on Intelligent Information Systems, Augustow, Poland, June 5-9, 1995.

Wnek, J. and Michalski, R.S., "Conceptual Transition from Logic to Arithmetic," Reports of Machine Learning and Inference Laboratory, MLI 94-7, Center for MLI, George Mason University, Fairfax, VA, December 1994.

Wnek, J. and Michalski, R.S., "Discovering Representation Space Transformations for Learning Concept Descriptions Combining DNF and M-of-N Rules," Working Notes of the ML-COLT'94 Workshop on Constructive Induction and Change of Representation, New Brunswick, NJ, July 1994.

Bloedorn, E., Michalski, R.S., and Wnek, J.,"Matching Methods with Problems: A Comparative Analysis of Constructive Induction Approaches," Reports of the Machine Learning and Inference Laboratory, MLI94-12, George Mason University, Fairfax, VA, 1994.

Arciszewski, E. Bloedorn, R.S. Michalski, M. Mustafa, J. Wnek, "Machine Learning in Conceptual Design: A Case Study on the Automated Acquisistion of Design Rules for Wind Bracings in Tall Buildings Using Constructive Inductive Learning," ASCE Journal of Computing in CE, Vol. 8, No. 3. July 1994. George Mason University, Fairfax, VA, 1994.

Wnek, J. and Michalski, R.S., "Hypothesis-driven Constructive Induction in AQ17-HCI: A Method and Experiments," Machine Learning, Vol. 14, No. 2, pp. 139-168, 1994.

Wnek, J., "Hypothesis-driven Constructive Induction," Ph.D. dissertation, School of Information Technology and Engineering, Reports of Machine Learning and Inference Laboratory, MLI 93-2, Center for Artificial Intelligence, George Mason University, (also published by University Microfilms Int., Ann Arbor, MI), March 1993.

For more references, see Publication section.