A Framework for Learning Rules from Multiple Instance Data
Abstract
This paper proposes a generic extension to propositional rule learners to handle multiple-instance data. In a multiple-instance representation, each learning example is represented by a bag” of fixed-length feature vectors”.Such a representation, lying somewhere between propositional and first-order representation, offers a tradeoff between the two. N aive -R ipper M i is one implementation of this extension on the rule learning algorithm R ipper . Several pitfalls encountered by this naive extension during induction are explained. A new multiple-instance search bias based on decision tree techniques is then used to avoid these pitfalls. Experimental results show the benefits of this approach for solving propositionalized relational problems in terms of speed and accuracy.
Cite
Text
Chevaleyre and Zucker. "A Framework for Learning Rules from Multiple Instance Data." European Conference on Machine Learning, 2001. doi:10.1007/3-540-44795-4_5Markdown
[Chevaleyre and Zucker. "A Framework for Learning Rules from Multiple Instance Data." European Conference on Machine Learning, 2001.](https://mlanthology.org/ecmlpkdd/2001/chevaleyre2001ecml-framework/) doi:10.1007/3-540-44795-4_5BibTeX
@inproceedings{chevaleyre2001ecml-framework,
title = {{A Framework for Learning Rules from Multiple Instance Data}},
author = {Chevaleyre, Yann and Zucker, Jean-Daniel},
booktitle = {European Conference on Machine Learning},
year = {2001},
pages = {49-60},
doi = {10.1007/3-540-44795-4_5},
url = {https://mlanthology.org/ecmlpkdd/2001/chevaleyre2001ecml-framework/}
}