Isolated hand-written digit recognition using a neurofuzzy scheme and multiple classification

摘要: A neuro-fuzzy system for isolated hand-written digit recognition using a similarity fuzzy measure is presented. The composed of two main blocks: first block that normalises the input and compares it with set patterns, second multilayer perceptron (MLP) to perform definitive classification. comparison patterns carried out via uses Yager parametric norms co-norms. Along this work, several values parameter have been studied, in order obtain optimum. simplicity method makes extremely quick. Recognition accuracy about 90% single classification, close 97,5% classification scheme.