Applying A Hybrid Method To Handwritten Character Recognition
Fu Chang, Chin-Chin Lin and Chun-Jen Chen Institute of Information Science, Academia Sinica, Taipei, Taiwan Dept. of Electrical Engineering, National Taipei University of Technology, Taipei, Taiwan E-mail: fchang@iis.sinica.edu.tw, erikson@iis.sinica.edu.tw, dean@iis.sinica.edu.tw
Abstract
In this paper, 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.
sample or k-nearest samples as the basis for classification, takes no training time and is usually faster than SVM in large-scale pattern matching applications. In practice, the NN method is still too slow and does not perform as well as SVM. We propose a method that exploits the advantages of both NN and SVM, and avoids their deficiencies. We conduct our method as follows. In the matching process, the set of all training samples is replaced by a much smaller set of prototypes. The SVM method is then used in a post-process that works on the top-rank candidates that have been obtained in the prototype-matching process. This paper presents all the ingredients in the training and testing processes associated with the hybrid method.

下一页