Investigation on the architecture of a fuzzy-neural computing system

In addition to the well-known massively parallel distributed processing features, the main reason why the brain system of human beings is more intelligent and effective than traditional computers of Von Neumann type is that it is able to process various types of fuzzy information, and moreover, instead of a precise solution, it only seeks for some satisfactory solutions. In this paper a class of neural computing system, the inputs of which are of fuzzy type and the outputs of which are satisfactory solutions is suggested. The architecture of the fuzzy neural computing system is outlined, some new results concerning fuzzy associative memory are also given.<>