Genetic Translator: How to Apply Query Learning to Practical OCR

We propose a novel learning method combining query learning and a “genetic translator” we developed. Query learning is a useful technique for high-accuracy, high-speed learning. However, it has not been applied for practical optical character readers (OCRs), since human beings cannot recognize queries in the feature space used in practical OCR devices. We previously proposed a character image reconstruction method using the Genetic Algorithm. Here, this method is applied as a “translator” from feature space for query learning of character recognition. The results of an experiment with hand-printed numeral recognition demonstrate the potential of the proposed method.