Presentation 2002/10/10
Class-Specific Feature Polynomial Classifier for Handwriting Recognition
Cheng-Lin Liu, Hiroshi Sako,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In handwriting recognition integrating segmentation and classification, the underlying classifier is desired to provide both high classification accuracy and resistance to out-of-class patterns (outliers). To achieve this goal, this paper proposes a class-specific feature polynomial classifier (CFPC), which can be viewed as a hybrid of ordinary polynomial classifier (PC) and projection distance method. The class-specific features are extracted by K-L transform on class-specific space, unlike that the ordinary PC uses the principal components on the whole pattern space. The incorporation of class-specific residual of subspace projection (projection distance) into CFPC was shown to benefit both classification accuracy and outlier resistance. The connecting weights of CFPC are learned class-by-class to minimize the mean square error on a training dataset. In experiments of handwritten digit recognition and numeral string recognition, the performance of CFPC is comparable or better than those of ordinary PC, learning quadratic discriminant function (LQDF), and support vector machines (SVMs).
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Pattern classification / Outlier resistance / Class-specific feature polynomial classifier / Handwritten digit recognition / Numeral string recognition
Paper # NC2002-44
Date of Issue

Conference Information
Committee NC
Conference Date 2002/10/10(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Class-Specific Feature Polynomial Classifier for Handwriting Recognition
Sub Title (in English)
Keyword(1) Pattern classification
Keyword(2) Outlier resistance
Keyword(3) Class-specific feature polynomial classifier
Keyword(4) Handwritten digit recognition
Keyword(5) Numeral string recognition
1st Author's Name Cheng-Lin Liu
1st Author's Affiliation Central Research Laboratory, Hitachi, Ltd.()
2nd Author's Name Hiroshi Sako
2nd Author's Affiliation Central Research Laboratory, Hitachi, Ltd.
Date 2002/10/10
Paper # NC2002-44
Volume (vol) vol.102
Number (no) 381
Page pp.pp.-
#Pages 6
Date of Issue