A class-modular GLVQ ensemble with outlier learning for handwritten digit recognition

Authors:
Katsuhiko Takahashi;Daisuke Nishiwaki
Affiliations:
-;-
Venue:
ICDAR '03 Proceedings of the Seventh International Conference on Document Analysis and Recognition - Volume 1
Year:
2003

Citing 2
Cited 2

Sorting and Recognizing Cheques and Financial Documents

DAS '98 Selected Papers from the Third IAPR Workshop on Document Analysis Systems: Theory and Practice
Improving rejection performance on handwritten digits by training with “rubbish”

Neural Computation

Histogram clustering and hybrid classifier for handwritten Arabic characters recognition

SPPRA'06 Proceedings of the 24th IASTED international conference on Signal processing, pattern recognition, and applications
Rapid and brief communication: Initialization insensitive LVQ algorithm based on cost-function adaptation

Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

A class-modular generalized learning vector quantization(GLVQ) ensemble method with outlier learning forhandwritten digit recognition is proposed. A GLVQ classifieris one of discriminative methods. Though discriminativeclassifiers have remarkable ability to solve characterrecognition problems, they are poor at outlier resistance. Toovercome this problem, a GLVQ classifier trained with bothdigit images and outlier images is introduced. Moreover,the original 10-classification problem is separated into ten2-classification problems using ten GLVQ classifiers, eachof which recognizes its corresponding digit class. Experimentalresults of handwritten digit recognition and outlierrejection reveal that our method is far more superior at outlierresistance than a conventional GLVQ classifier, whilemaintaining its digit recognition performance.