DocumentCode :
419626
Title :
Applying a hybrid method to handwritten character recognition
Author :
Chang, Fu ; Lin, Chin-Chin ; Chen, Chun-Jen
Author_Institution :
Inst. of Information Sci., Acad. Sinica, Taipei, Taiwan
Volume :
2
fYear :
2004
fDate :
23-26 Aug. 2004
Firstpage :
529
Abstract :
We propose a new prototype learning/matching method that can be combined with support vector machines (SVM) in pattern recognition. This hybrid method has the following merits. One, the learning algorithm for constructing prototypes determines both the number and the location of prototypes. This algorithm terminates within a finite number of iterations and assures that each training sample matches in class types with the nearest prototype. Two, SVM can be used to process top-rank candidates obtained by the prototype learning/matching method so as to save time in both training and testing processes. We apply our method to recognizing handwritten numerals and handwritten Chinese/Hiragana characters. Experiment results show that the hybrid method saves great amount of training and testing time in large-scale tasks and achieves comparable accuracy rates to those achieved by using SVM solely. Our results also show that the hybrid method performs better than the nearest neighbour method.
Keywords :
handwritten character recognition; learning (artificial intelligence); natural languages; support vector machines; handwritten Chinese character recognition; handwritten Hiragana character recognition; handwritten character recognition; handwritten numerals recognition; pattern recognition; prototype learning method; prototype matching method; support vector machines; Character recognition; Information science; Large-scale systems; Neural networks; Pattern matching; Pattern recognition; Prototypes; Support vector machine classification; Support vector machines; Testing;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
ISSN :
1051-4651
Print_ISBN :
0-7695-2128-2
Type :
conf
DOI :
10.1109/ICPR.2004.1334291
Filename :
1334291
Link To Document :
بازگشت