Idiotypic Immune Network Based Multiagent

In many years, natural including human being, which is important content and orientation in scientific research, are always interested by scientists and engineers. Via researching in this field, we master a lot of natural information which was mystical. It is more important that we can using these mysteries continue our way of cognising nature conquering nature and harmonious coexisting with nature.The conception of Agent came from the cognition of artificial intelligence: the final aim of artificial intelligence is achieving intelligent "agent" that can replace human transacting affairs. So, during the early phase, people devoted a lot of energy to the research of mental state of Agent. A great deal of models and theories were engendered one after the other, for instance, BDI model by Cohen-LevesqueN BDI model by Rao-Georgeff and so on. But along with the deep research to Agent, the cognition of Agent have been extensive. Agent have become a method to describing intricate phenomenon researching intricate systems, achieving intricate self-adaptive calculate.Based on idiotypic network theory and Multiagent system theory, a novel artificial immune system model, idiotypic network based on Multiagent and corresponding algorithm, are proposed. The antibodies produced by the B cells are regard as agents in the model. They have local perceptivity, the abilities of competition, cooperation, self-learning etc. A group of agents belonged to the same B cell is a solution in solution space. After competition, cooperation among agents and development of network, the individuals in network include satisfactory solution, and then the problem is solved. In experiments, we apply the proposed algorithm to TSP Problem, and the results shown to be capable of solving complex machine learning tasks effectively.