Exponentiated Gradient Exploration for Active Learning

Active learning strategies respond to the costly labeling task in a supervised classification by selecting the most useful unlabeled examples in training a predictive model. Many conventional active learning algorithms focus on refining the decision boundary, rather than exploring new regions that c...

Full description

Bibliographic Details
Main Author:	Djallel Bouneffouf
Format:	Article
Language:	English
Published:	MDPI AG 2016-01-01
Series:	Computers
Subjects:	active learning exploration and exploitation exponentiated gradient
Online Access:	http://www.mdpi.com/2073-431X/5/1/1

id	doaj-547458fa30094c2cad4bf8316cba0b9a
record_format	Article
spelling	doaj-547458fa30094c2cad4bf8316cba0b9a2020-11-24T23:30:48ZengMDPI AGComputers2073-431X2016-01-0151110.3390/computers5010001computers5010001Exponentiated Gradient Exploration for Active LearningDjallel Bouneffouf0Department of Computer Science, Télécom SudParis, UMR CNRS Samovar, 91011 Evry Cedex, FranceActive learning strategies respond to the costly labeling task in a supervised classification by selecting the most useful unlabeled examples in training a predictive model. Many conventional active learning algorithms focus on refining the decision boundary, rather than exploring new regions that can be more informative. In this setting, we propose a sequential algorithm named exponentiated gradient (EG)-active that can improve any active learning algorithm by an optimal random exploration. Experimental results show a statistically-significant and appreciable improvement in the performance of our new approach over the existing active feedback methods.http://www.mdpi.com/2073-431X/5/1/1active learningexploration and exploitationexponentiated gradient
collection	DOAJ
language	English
format	Article
sources	DOAJ
author	Djallel Bouneffouf
spellingShingle	Djallel Bouneffouf Exponentiated Gradient Exploration for Active Learning Computers active learning exploration and exploitation exponentiated gradient
author_facet	Djallel Bouneffouf
author_sort	Djallel Bouneffouf
title	Exponentiated Gradient Exploration for Active Learning
title_short	Exponentiated Gradient Exploration for Active Learning
title_full	Exponentiated Gradient Exploration for Active Learning
title_fullStr	Exponentiated Gradient Exploration for Active Learning
title_full_unstemmed	Exponentiated Gradient Exploration for Active Learning
title_sort	exponentiated gradient exploration for active learning
publisher	MDPI AG
series	Computers
issn	2073-431X
publishDate	2016-01-01
description	Active learning strategies respond to the costly labeling task in a supervised classification by selecting the most useful unlabeled examples in training a predictive model. Many conventional active learning algorithms focus on refining the decision boundary, rather than exploring new regions that can be more informative. In this setting, we propose a sequential algorithm named exponentiated gradient (EG)-active that can improve any active learning algorithm by an optimal random exploration. Experimental results show a statistically-significant and appreciable improvement in the performance of our new approach over the existing active feedback methods.
topic	active learning exploration and exploitation exponentiated gradient
url	http://www.mdpi.com/2073-431X/5/1/1
work_keys_str_mv	AT djallelbouneffouf exponentiatedgradientexplorationforactivelearning
_version_	1725540242618843136

Exponentiated Gradient Exploration for Active Learning

Similar Items