New boosting algorithms for classification problems with large number of classes applied to a handwritten word recognition task. (English) Zbl 1040.68642

Günter, Simon; Bunke, Horst

Windeatt, Terry (ed.) et al., Multiple classifier systems. 4th international workshop, MCS 2003, Guildford, UK, June 11–13, 2003. Proceedings. Berlin: Springer (ISBN 3-540-40369-8/pbk). Lect. Notes Comput. Sci. 2709, 326-335 (2003).

Summary: Methods that create several classifiers out of one base classifier, so-called ensemble creation methods, have been proposed and successfully applied to many classification problems recently. One category of such methods is Boosting with AdaBoost being the best known procedure belonging to this category. Boosting algorithms were first developed for two-class problems, but then extended to deal with multiple classes. Yet these extensions are not always suitable for problems with a large number of classes. In this paper we introduce some novel boosting algorithms which are designed for such problems, and we test their performance in a handwritten word recognition task.
For the entire collection see [Zbl 1029.68928].

MSC:

68U99	Computing methodologies and applications
68T10	Pattern recognition, speech recognition
68T05	Learning and adaptive systems in artificial intelligence

Keywords:

Multiple Classifier System; Ensemble Creation Method; Boosting; Large Number of Classes; Hidden Markov Model (HMM); Handwriting Recognition

Software:

IAM