Abstract
Expert classification systems have proven themselves effective decision makers for many types of problems. However, the accuracy of such systems is often highly dependent upon the accuracy of a human expert's domain theory. When human experts learn or create a set of rules, they are subject to a number of hindrances. Most significantly experts are, to a greater or lesser extent, restricted by the tradition of scholarship which has preceded them and by an inability to examine large amounts of data in a rigorous fashion without the effects of boredom or frustration. As a result, human theories are often erroneous or incomplete. To escape this dependency, machine learning systems have been developed to automatically refine and correct an expert's domain theory. When theory revision systems are applied to expert theories, they often concentrate on the reformulation of the knowledge provided rather than on the reformulation or selection of input features. The general assumption seems to be that the expert has already selected the set of features that will be most useful for the given task. That set may, however, be suboptimal. This paper studies theory refinement and the relative benefits of applying feature selection versus more extensive theory reformulation.
Article PDF
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
References
Almuallim, H. & Dietterich, T. G. (1991). Learning with many irrelevant features. The Proceedings of the Ninth National Conference on Artificial Intelligence (pp. 547–552).
Anglano, C., Giordana, A., Lo Bello, G., & Saitta, L. (1998). An experimental evaluation of coevolutive concept learning. The Proceedings of the Fifteenth International Conference on Machine Learning.
Asker, L. & Maclin, R. (1997). Feature engineering and classifier selection: A case study in venusian volcano detection. The Proceedings of the Fourteenth International Conference on Machine Learning.
Baffes, P. & Mooney, R. (1993). Extending theory revision to m-of-n rules. Informatica, 17(4), 387–397.
Bloedorn, E. & Michalski, R. (1998). Data-driven constructive induction. IEEE Trans. on Intelligent Systems, 13(2), 30–37.
Botta, M., Giordana, A., & Piolam, R. (1997). FONN: Combining first order logic with connectionist learning. The Proceedings of the Fourteenth International Conference on Machine Learning.
Botta, M. & Piolam, R. (1998). Refining numerical terms in structured FOL theories. The Proceedings of the Fourth International Workshop on Multistrategy Learning.
Breiman, L. (1996). Bagging predictors. Machine Learning, 24(2), 123–140.
Das, S., Giles, C. L., & Sun, G. (1992). Using prior knowledge in an nnpda to learn context-free languages. Advances in Neural Information Processing Systems (Vol. 5).
Devaney, M. & Ram, A. (1997). Efficient feature selection in conceptual clustering. The Proceedings of the Fourteenth International Conference on Machine Learning.
Frean, M. (1990). The upstart algorithm: A method for constructing and training feedforward neural networks. Neural Computation, 2(2), 189–209.
Kira, K. & Rendell, L. (1992). The feature selection problem: Traditional methods and a new algorithm. The Proceedings of the Tenth National Conference on Artificial Intelligence.
Maclin, R. & Shavlik, J. (1993). Using knowledge-based neural networks to improve algorithms: Refining the Chou-Fasman algorithm for protein folding. Machine Learning, 11, 195–215.
Mahoney, J. (1996). Combining symbolic and connectionist learning methods to refine certaintty-factor rule-bases. Ph.D. Thesis, University of Texas, Austin.
Omlin, C. & Giles, C. L. (1996). Rule revision with recurrent neural networks. IEEE Trans. on Knowledge and Data Engineering, 8(1), 183.
Opitz, D. W. (1995). An anytime approach to connectionist theory refinement: Refining the topologies of knowledge-based neural networks. Ph.D. thesis, Department of Computer Science, University of Wisconsin-Madison.
Opitz, D. W. & Shavlik, J. W. (1993). Heuristically expanding knowledge-based neural networks. Proceedings of the Thirteenth International Joint Conference on Artificial Intelligence (pp. 1360–1365).
Opitz, D. W. & Shavlik, J. W. (1996). Generating accurate and diverse members of a neural-network ensemble. Advances in Neural Information Processing Systems (Vol. 8, pp. 535–543).
Ourston, D. (1991). Using explanation based and empirical methods on theory revision. Ph.D. Thesis, University of Texas, Austin.
Quinlan, J. (1986). Induction of decision trees. Machine Learning, 1, 81–106.
Towell, G. & Shavlik, J. W. (1992). Using symbolic learning to improve knowledge-based neural networks. Proceedings of the Tenth National Conference on Artificial Intelligence (pp. 177–182).
Towell, G., Shavlik, J., & Noordewier, M. (1990). Refinement of approximate domain theories by knowledge-based neural networks. Proceedings of the Eighth National Conference on Artificial Intelligence (pp. 861–866).
Vafaie, H. & DeJong, K. (1998). Feature space transformation using genetic algorithms. IEEE Trans. on Intelligent Systems, 13(2), 57–65.
Yang, J. & Honavar, V. (1998). Feature subset selection using a genetic algorithm. IEEE Trans. on Intelligent Systems, 13(2), 44–49.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Burns, B.D., Danyluk, A.P. Feature Selection vs Theory Reformulation: A Study of Genetic Refinement of Knowledge-based Neural Networks. Machine Learning 38, 89–107 (2000). https://doi.org/10.1023/A:1007634023329
Issue Date:
DOI: https://doi.org/10.1023/A:1007634023329