Was the Year 2000 a Leap Year? Step-Wise Narrowing Theories with Metagol
- 508 Downloads
Many people believe that every fourth year is a leap year. However, this rule is too general: year X is a leap year if X is divisible by 4 except if X is divisible by 100 except if X is divisible by 400. We call such a theory with alternating generalisation and specialisation a step-wise narrowed theory. We present and evaluate an extension to the ILP system Metagol which facilitates learning such theories. We enabled Metagol to learn over-general theories by allowing a limited number of false positives during learning. This variant is iteratively applied on a learning task. For each iteration after the first, positive examples are the false positives from the previous iteration and negative examples are the true positives from the previous iteration. Iteration continues until no more false positives are present. Then, the theories are combined to a single step-wise narrowed theory. We evaluate the usefulness of our approach in the leap year domain. We can show that our approach finds solutions with fewer clauses, higher accuracy, and in shorter time.
KeywordsStep-wise narrowed theory Metagol Over-generalization
We like to thank Andrew Cropper for valuable discussions on negation in Metagol. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – SCHM 1239/10-1.
- 1.Richards, E.G.: Calendars. In: Urban, S.E., Seidelmann, P.K. (eds.) Explanatory Supplement to the Astronomical Almanac, pp. 585–624 (2013)Google Scholar
- 2.Quinlan, J.R.: Induction of decision trees. Mach. Learn. 1, 81–106 (1986)Google Scholar
- 4.Muggleton, S., Buntine, W.: Machine invention of first-order predicates by inverting resolution. In: Machine Learning Proceeding, pp. 339–352 (1988), https://doi.org/10.1016/B978-0-934613-64-4.50040-2
- 5.Bain, M., Muggleton, S.: Non-monotonic Learning. In: Machine Intelligence 12 - Towards an Automated Logic of Human Thought, pp. 105–120 (1991)Google Scholar
- 6.Malerba, D., Esposito, F., Lisi, F.A.: Learning Recursive Theories with ATRE. In: ECAI (European Conference on Artificial Intelligence), pp. 435–439 (1998)Google Scholar
- 7.Malerba, D.: Learning recursive theories in the normal ILP setting. Fundamenta Informaticae 57, 39–77 (2003)Google Scholar
- 9.Stahl, I.: Predicate Invention in ILP - an Overview. In: Machine Learning: ECML-1993, pp. 313–322 (1993)Google Scholar
- 11.Lin, D., Dechter, E., Ellis, K., Tenenbaum, J., Muggleton, S.H.: Bias reformulation for one-shot function induction. In: ECAI (European Conference on Artificial Intelligence), pp. 525–530 (2014). https://doi.org/10.3233/978-1-61499-419-0-525
- 12.Cropper, A., Muggleton, S.H.: Learning higher-order logic programs through abstraction and invention. In: IJCAI (International Joint Conference on Artificial Intelligence), pp. 1418–1424 (2016)Google Scholar
- 14.Cropper, A., Muggleton, S.H.: Metagol System (2016). https://github.com/metagol/metagol
- 15.Larson, J., Michalski, R.S.: Inductive Inference of VL Decision Rules. ACM SIGART Bull., 38–44, June 1977Google Scholar