An autonomous decentralized recognition system having a dispersive wave property

The concept of an autonomous decentralized system can be applied to recognizing various complicated and skillful actions performed by humans. This concept is the basis for a proposed pattern recognition system, which can be regarded as an associative memory system modeled by a reaction-diffusion equation system. This paper improves upon this pattern recognition system using the property of a dispersive wave instead of diffusion. This improvement has three advantages: acceleration of recognition, spread of similar regions and human-like hysteresis in recognition. Computer simulation has shown that propagation of a dispersive wave is faster than that of diffusion, and that the dispersive wave system exhibits the above advantageous properties of recognition.