Ensemble pruning using reinforcement learning

Ioannis Partalas, Grigorios Tsoumakas, Ioannis Katakis, Ioannis Vlahavas

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Multiple Classifier systems have been developed in order to improve classification accuracy using methodologies for effective classifier combination. Classical approaches use heuristics, statistical tests, or a meta-learning level in order to find out the optimal combination function. We study this problem from a Reinforcement Learning perspective. In our modeling, an agent tries to learn the best policy for selecting classifiers by exploring a state space and considering a future cumulative reward from the environment. We evaluate our approach by comparing with state-of-the-art combination methods and obtain very promising results.

Original languageEnglish
Title of host publicationAdvances in Artificial Intelligence - 4th Helenic Conference on AI, SETN 2006, Proceedings
PublisherSpringer Verlag
Pages301-310
Number of pages10
ISBN (Print)354034117X, 9783540341178
DOIs
Publication statusPublished - 1 Jan 2006
Event4th Helenic Conference on AI, SETN 2006 - Heraklion, Crete, Greece
Duration: 18 May 200620 May 2006

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3955 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference4th Helenic Conference on AI, SETN 2006
Country/TerritoryGreece
CityHeraklion, Crete
Period18/05/0620/05/06

Fingerprint

Dive into the research topics of 'Ensemble pruning using reinforcement learning'. Together they form a unique fingerprint.

Cite this