Active Learning for Parzen Window Classifier
Abstract
The problem of active learning is approached in this paper by minimizing\ndirectly an estimate of the expected test error. The main difficulty\nin this ``optimal'' strategy is that output probabilities need to be \nestimated accurately. We suggest here different methods\nfor estimating those efficiently.\nIn this context, the Parzen window classifier is considered\nbecause it is both simple and probabilistic. The analysis of experimental\nresults highlights that regularization is a key ingredient for this strategy.
Cite
Text
Chapelle. "Active Learning for Parzen Window Classifier." Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics, 2005.Markdown
[Chapelle. "Active Learning for Parzen Window Classifier." Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics, 2005.](https://mlanthology.org/aistats/2005/chapelle2005aistats-active/)BibTeX
@inproceedings{chapelle2005aistats-active,
title = {{Active Learning for Parzen Window Classifier}},
author = {Chapelle, Olivier},
booktitle = {Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics},
year = {2005},
pages = {49-56},
volume = {R5},
url = {https://mlanthology.org/aistats/2005/chapelle2005aistats-active/}
}