Combination of statistical and neural classifiers for a high‐accuracy recognition of large character sets

Abstract

SUMMARYIn this paper the authors propose a method for highaccuracy recognition of large character sets using a new combination of a statistical method and neural networks. In their method, a hierarchical structure that has several neural networks arranged in a line after the statistical method is used. First, recognition using a statistical method is performed, and this represents the final result if the top candidate does not belong to a predefined set of similar characters. If it does, then the input character is discriminated in a neural network which designates the top candidate by determining the similar characters. The results are output as final results. The basic idea of this method is the functional division of a statistical method and neural networks, and the use of a neural network as determined by a statistical method. The results of recognizing 3201 character types including JIS-1 Kanji showed an improvement in the correct recognition rate due to the combined use of a statistical method and neural networks, thereby demonstrating the validity of the authors' approach.

Error Reduction by Confusing Characters Discrimination for Online Handwritten Japanese Character Recognition

“…Compared to the CMF, the proposed method trains discriminative confusion set classifiers instead of generative classifiers. Compared to the pre-trained pair discriminators of [6] [7] and the neural networks in [8], the proposed method costs only small storage of extra parameters. Our experimental results on online handwritten Japanese characters show that the proposed method largely reduces the recognition errors of non-Kanji characters.…”

“…A more straightforward strategy for discriminating confusing characters is just to re-classify the input pattern in a subset of confusing classes with a discriminative classifier, such as the neural network in [8]. This method re-classifies the input pattern whenever the top rank class of baseline classifier falls in a confusion set.…”