Improving Sequential Feature Selection Methods Performance by Means of Hybridization

0341554 - ÚTIA 2011 RIV CA eng C - Conference Paper (international conference)
Somol, Petr - Novovičová, Jana - Pudil, Pavel
Improving Sequential Feature Selection Methods Performance by Means of Hybridization.
Proc. 6th IASTED Int. Conf. on Advances in Computer Science and Engineering. Calgary: ACTA Press, 2010 - (Rafea), 689-1-689-10. ISBN 978-0-88986-830-4.
[Advances in Computer Science and Engineering. Sharm El Sheikh (EG), 15.03.2010-17.03.2010]
R&D Projects: GA MŠMT 1M0572; GA ČR GA102/08/0593; GA ČR GA102/07/1594
Grant - others:GA MŠk(CZ) 2C06019
Institutional research plan: CEZ:AV0Z10750506
Keywords : Feature selection * sequential search * hybrid methods * classification performance * subset search * statistical pattern recognition
Subject RIV: BD - Theory of Information
http://library.utia.cas.cz/separaty/2010/RO/somol-improving sequential feature selection methods performance by means of hybridization.pdf

In this paper we propose the general scheme of defining hybrid feature selection algorithms based on standard sequential search with the aim to improve feature selection performance, especially on high-dimensional or large-sample data. We show experimentally that “hybridization” has not only the potential to dramatically reduce FS search time, but in some cases also to actually improve classifier generalization, i.e., its classification performance on previously unknown data.
Permanent Link: http://hdl.handle.net/11104/0184495