A Neural Network Model for Pattern Recognition

摘要

We have developed a small scale four-layered neural network (NN) model for simple character recognition, which can recognize the patterns transformed by affine conversion. It learns by back-propagation to obtain the characteristics of the simple cell and the complex cell in the visual cortex. In this study 24 patterns are presented as input patterns. An input pattern is divided into 64 local patterns and connected with the 1st hidden layer as in the visual cortex. The proposed NN model has good performance of the feature extraction in first layers.