A Smallest Generalization Step Strategy
Abstract
Over-generalization is a well-known problem in empirical learning. Incremental and prudent generalization is a means to avoid it. This is not always sufficient. The language in which the concepts are described may be incomplete, so that there is no conjunction to express a concept that is consistent with all the examples. This paper presents an interactive incremental learning method that generalizes in such a way that it is able to efficiently assist an user in locating the insufficiencies of the language and in correcting them whenever an over-generalization occurs. The generalization algorithm is based on a smallest generalization step strategy that determines processing order of the examples and the successive hypothesis to study.
Cite
Text
Nedellec. "A Smallest Generalization Step Strategy." International Conference on Machine Learning, 1991. doi:10.1016/B978-1-55860-200-7.50108-2Markdown
[Nedellec. "A Smallest Generalization Step Strategy." International Conference on Machine Learning, 1991.](https://mlanthology.org/icml/1991/nedellec1991icml-smallest/) doi:10.1016/B978-1-55860-200-7.50108-2BibTeX
@inproceedings{nedellec1991icml-smallest,
title = {{A Smallest Generalization Step Strategy}},
author = {Nedellec, Claire},
booktitle = {International Conference on Machine Learning},
year = {1991},
pages = {529-533},
doi = {10.1016/B978-1-55860-200-7.50108-2},
url = {https://mlanthology.org/icml/1991/nedellec1991icml-smallest/}
}