Soft computing (immune networks) in artificial intelligence

This paper proposes a novel reactive distributed artificial intelligence (dynamic) using immune networks and other soft computing methods. Firstly, extended soft computing is defined by adding immune networks and chaos theory including fractal and wavelet to conventional soft computing which is the fusion or combination of fuzzy systems, neural networks and genetic algorithms and is suitable for cognitive distributed artificial intelligence (static). Next, a novel fuzzy neural net (general parameter radial based function neural network) is developed in order to use it for communication among agents in immune networks. The general parameter method is extended to an adaptive structured genetic algorithm to obtain a much faster convergence rate. An unbiasedness criterion using distorter (a kind of GMDH) is applied to better generalization properties. Then, this developed fuzzy neural net is extended to a high performance radial based function network in order to optimize parameters resulting in the reactive distributed artificial intelligence