DocumentCode :
2474235
Title :
Incremental learning in non-stationary environments with concept drift using a multiple classifier based approach
Author :
Karnick, Matthew ; Muhlbaier, Michael D. ; Polikar, Robi
Author_Institution :
Electr. & Comput. Eng., Rowan Univ., Glassboro, NJ, USA
fYear :
2008
fDate :
8-11 Dec. 2008
Firstpage :
1
Lastpage :
4
Abstract :
We outline an incremental learning algorithm designed for nonstationary environments where the underlying data distribution changes over time. With each dataset drawn from a new environment, we generate a new classifier. Classifiers are combined through dynamically weighted majority voting, where voting weights are determined based on classifiers¿ age and accuracy on current and past environments. The most recent and relevant classifiers are weighted higher, allowing the algorithm to appropriately adapt to drifting concepts. This algorithm does not discard prior classifiers, allowing efficient learning of potentially cyclical environments. The algorithm learns incrementally, i.e., without access to previous data. Finally, the algorithm can use any supervised classifier as its base model, including those not normally capable of incremental learning. We present the algorithm and its performance using different base learners in different environments with varying types of drift.
Keywords :
learning (artificial intelligence); pattern classification; drifting concept; dynamically weighted majority voting; incremental learning; multiple classifier; nonstationary environment; Algorithm design and analysis; Boosting; Change detection algorithms; Data engineering; Design engineering; Distributed computing; Nonlinear equations; Pattern recognition; Training data; Voting;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Pattern Recognition, 2008. ICPR 2008. 19th International Conference on
Conference_Location :
Tampa, FL
ISSN :
1051-4651
Print_ISBN :
978-1-4244-2174-9
Electronic_ISBN :
1051-4651
Type :
conf
DOI :
10.1109/ICPR.2008.4761062
Filename :
4761062
Link To Document :
بازگشت