Learnability of Constrained Logic Programs
Abstract
The field of Inductive Logic Programming (ILP) is concerned with inducing logic programs from examples in the presence of back-ground knowledge. This paper defines the ILP problem and describes several syntactic restrictions that are often used in ILP. We then derive some positive results concerning the learnability of these restricted classes of logic programs, by reduction to a standard propositional learning problem. More specifically, k -literal predicate definitions consisting of constrained, function-free, nonrecursive program clauses are polynomially PAC-learnable under arbitrary distributions.
Cite
Text
Dzeroski et al. "Learnability of Constrained Logic Programs." European Conference on Machine Learning, 1993. doi:10.1007/3-540-56602-3_148Markdown
[Dzeroski et al. "Learnability of Constrained Logic Programs." European Conference on Machine Learning, 1993.](https://mlanthology.org/ecmlpkdd/1993/dzeroski1993ecml-learnability/) doi:10.1007/3-540-56602-3_148BibTeX
@inproceedings{dzeroski1993ecml-learnability,
title = {{Learnability of Constrained Logic Programs}},
author = {Dzeroski, Saso and Muggleton, Stephen H. and Russell, Stuart},
booktitle = {European Conference on Machine Learning},
year = {1993},
pages = {342-347},
doi = {10.1007/3-540-56602-3_148},
url = {https://mlanthology.org/ecmlpkdd/1993/dzeroski1993ecml-learnability/}
}