Position independent pattern matching by neural network

A novel pattern-matching neural network is proposed. The network matches an input to multiple candidates of the stored templates in parallel. It can find the best matching template, whose features are arranged in the same order as those of the input, regardless of positional differences between corresponding features. The network consists of a pattern-matching layer, a minimal distance detection layer, and a recognition layer. The matching between an input pattern and candidates of the to-be-matched template is performed in the pattern-matching layer. The candidates are selected in the minimal distance detection layer, and are fed back in parallel to the pattern-matching layer. The candidate that has features in the same positional order as the input can be determined in the pattern-matching layer. The matched template is classified by the recognition layer. Results of simulation studies on one- and two-dimensional patterns have shown that the network can perform position-independent pattern matching completely