Learning Active Classifiers
Abstract
Many classification algorithms are "passive", in that they assign a class-label to each instance based only on the description given, even if that description is incomplete. In contrast, an active classifier can --- at some cost --- obtain the values of missing attributes, before deciding upon a class label. The expected utility of using an active classifier depends on both the cost required to obtain the additional attribute values and the penalty incurred if it outputs the wrong classification. This paper considers the problem of learning near-optimal active classifiers, using a variant of the probably-approximatelycorrect (PAC) model. After defining the framework --- which is perhaps the main contribution of this paper --- we describe a situation where this task can be achieved efficiently, but then show that the task is often intractable. Appears in the Proceedings of the Thirteenth International Conference on Machine Learning (IMLC-96), Bari Italy, July 1996. 1 INTRODUCTION A ...
Cite
Text
Greiner et al. "Learning Active Classifiers." International Conference on Machine Learning, 1996.Markdown
[Greiner et al. "Learning Active Classifiers." International Conference on Machine Learning, 1996.](https://mlanthology.org/icml/1996/greiner1996icml-learning/)BibTeX
@inproceedings{greiner1996icml-learning,
title = {{Learning Active Classifiers}},
author = {Greiner, Russell and Grove, Adam J. and Roth, Dan},
booktitle = {International Conference on Machine Learning},
year = {1996},
pages = {207-215},
url = {https://mlanthology.org/icml/1996/greiner1996icml-learning/}
}