bibtype C - Conference Paper (international conference)
ARLID 0348710
utime 20240111140745.8
mtime 20101101235959.9
title (primary) (eng) The Problem of Fragile Feature Subset Preference in Feature Selection Methods and A Proposal of Algorithmic Workaround
specification
page_count 4 s.
media_type flash
serial
ARLID cav_un_epca*0348709
ISBN 978-0-7695-4109-9
ISSN 1051-4651
title Proc. 2010 Int. Conf. on Pattern Recognition
page_num 4396-4399
publisher
place Istanbul
name IEEE Computer Society
year 2010
keyword feature selection
keyword machine learning
keyword over-fitting
keyword classification
keyword feature weights
keyword weighted features
keyword feature acquisition cost
author (primary)
ARLID cav_un_auth*0101197
name1 Somol
name2 Petr
full_dept (cz) Rozpoznávání obrazu
full_dept (eng) Department of Pattern Recognition
department (cz) RO
department (eng) RO
institution UTIA-B
full_dept Department of Pattern Recognition
fullinstit Ústav teorie informace a automatizace AV ČR, v. v. i.
author
ARLID cav_un_auth*0101091
name1 Grim
name2 Jiří
full_dept (cz) Rozpoznávání obrazu
full_dept Department of Pattern Recognition
department (cz) RO
department RO
institution UTIA-B
full_dept Department of Pattern Recognition
fullinstit Ústav teorie informace a automatizace AV ČR, v. v. i.
author
ARLID cav_un_auth*0101182
name1 Pudil
name2 Pavel
full_dept (cz) Rozpoznávání obrazu
full_dept Department of Pattern Recognition
department (cz) RO
department RO
institution UTIA-B
full_dept Department of Pattern Recognition
fullinstit Ústav teorie informace a automatizace AV ČR, v. v. i.
source
url http://library.utia.cas.cz/separaty/2010/RO/somol-the problem of fragile feature subset preference in feature selection methods and a proposal of algorithmic workaround.pdf
source_size 474kb
cas_special
project
project_id 2C06019
agency GA MŠk
country CZ
ARLID cav_un_auth*0216518
project
project_id GA102/07/1594
agency GA ČR
ARLID cav_un_auth*0228611
project
project_id GA102/08/0593
agency GA ČR
ARLID cav_un_auth*0239567
project
project_id 1M0572
agency GA MŠk
ARLID cav_un_auth*0001814
research CEZ:AV0Z10750506
abstract (eng) We point out a problem inherent in the optimization scheme of many popular feature selection methods. It follows from the implicit assumption that higher feature selection criterion value always indicates more preferable subset even if the value difference is marginal. This assumption ignores the reliability issues of particular feature preferences, overfitting and feature acquisition cost. We propose an algorithmic extension applicable to many standard feature selection methods allowing better control over feature subset preference. We show experimentally that the proposed mechanism is capable of reducing the size of selected subsets as well as improving classifier generalization.
action
ARLID cav_un_auth*0265069
name 20th International Conference on Pattern Recognition
place Istanbul
dates 23.08.2010-26.08.2010
country TR
reportyear 2011
RIV BD
permalink http://hdl.handle.net/11104/0189152
mrcbT16-q 50
mrcbT16-s 0.420
mrcbT16-y 10.52
mrcbT16-x 0.85
arlyear 2010
mrcbU56 474kb
mrcbU63 cav_un_epca*0348709 Proc. 2010 Int. Conf. on Pattern Recognition 978-0-7695-4109-9 1051-4651 4396 4399 Istanbul IEEE Computer Society 2010