Patterns and symbols for solving ill-defined problems

The paper presents a novel network for inference using patterns and symbols. The meanings of patterns and symbols are also given. The network consists of a self organizing network and a two layered associative memory network having a pattern layer and a symbol layer. The proposed network is an implementation of a new framework for solving ill defined problems using patterns and symbols. Candidates for solution of inputted image are generated by the network through associations from a pattern to another pattern, from a symbol to another symbol, and inference from patterns to symbols and vice versa.