DocumentCode :
2118146
Title :
Online random forests based on CorrFS and CorrBE
Author :
Elgawi, O.H.
Author_Institution :
Tokyo Inst. of Technol., Tokyo
fYear :
2008
fDate :
23-28 June 2008
Firstpage :
1
Lastpage :
7
Abstract :
This paper aims to contribute to the merits of online ensemble learning for classification problems. To this end we induce random forests algorithm into online mode and estimate the importance of variables incrementally based on correlation ranking (CR). We test our method by an ldquoincremental hill climbingrdquo algorithm in which features are greedily added in a ldquoforwardrdquo step (FS), and removed in a ldquobackwardrdquo step (BE). We resort to an implementation that combine CR with FS and BE. We call this implementation CorrFS and CorrBE respectively. Evaluation based on public UCI databases demonstrates that our method can achieve comparable performance to classifiers constructed from batch training. In addition, the framework allows a fair comparison among other batch mode feature selection approaches such as Gini index, ReliefF and gain ratio.
Keywords :
learning (artificial intelligence); pattern classification; Gini index; ReliefF; classification problems; correlation ranking; gain ratio; incremental hill climbing; online ensemble learning; online random forests; Algorithm design and analysis; Bagging; Chromium; Decision trees; Input variables; Machine learning; Machine learning algorithms; Radio frequency; Spatial databases; Testing;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Vision and Pattern Recognition Workshops, 2008. CVPRW '08. IEEE Computer Society Conference on
Conference_Location :
Anchorage, AK
ISSN :
2160-7508
Print_ISBN :
978-1-4244-2339-2
Electronic_ISBN :
2160-7508
Type :
conf
DOI :
10.1109/CVPRW.2008.4563065
Filename :
4563065
Link To Document :
بازگشت