METHOD AND DEVICE FOR LEARNING PATTERN RECOGNITION DEVICE USING NEURAL NETWORK

摘要

PURPOSE:To efficientiy execute learning processing and to shorten processing time in a pattern recognition device using a neural network. CONSTITUTION:The small quantity of representative pattern data is extracted from a pattern data file 7 to be learned in each pattern sort, a representative pattern data file 9 constituted of the extracted pattern data and pattern data which can not be correctly recognized by the neural network 5 is used as a substitution for the original file 7 and leaned by the network 5 to compress data volume for learning processing. Processings for learning the network 5 by file 9, recognizing the original file 7, extracting pattern data which can not be correctly and first recognized, updating the file 9, are executed and learning again the network 5 by the updated file 9 is repeated.