The Applications Of Immune Algorithm And Kernel Clustering Artificial Immune Network

Because of its important role in our body and its outstanding properties, Natural immune system has allured a lot of attention. People not only research the mechanism of immune system by modern technology tools, but also design many new algorithms and tools imitating the mechanism of natural immune system to deal with many engineering problems, which are difficult for traditional tools to solve.However, Artificial immune system (AIS) is a new research area, many people devote to this work, and AIS becomes a rising hot point of evolutionary computation and pattern recognition, but, in the whole, the results are in disorder and short of systematization. Especially, some so-called AIS essentially are modified forms of traditional methods, not really systems based immune system. Furthermore, both modeling and application of AIS are not so thorough and perfect that the important characteristics of natural immune system are not fully realized and utilized.Based on these problems, researching work in this paper is mainly concentrated on modeling of AIS, design of its training algorithm, and definitude of its application.Begin with reading a lot of documents and thinking deeply, summarize the important and useful mechanisms of natural immune system, and the research works on AIS and its application with more systematic viewpoint, then divide existing AIS into three groups: immune algorithm, artificial immune network, and artificial immune system. Based on former works, put forward a common frame of AIS and the mainly application of AIS, which is character abstract. Then analyze the existing immune algorithms according to the common frame of AIS, and present a modified immune algorithm, called Multi-peaks Searching Immune Algorithm. By simulation experiments, the performance of the algorithm is studied, and the difference between Genetic Algorithm and this algorithm is discussed. Aim at clustering analysis, based on the application of AIS, present a new form of AIS, called Kernel Clustering Artificial Immune Network, also by simulation experiments, the performance of the AIS is discussed, and the validity of this AIS solving clustering analysis is showed.In a word, the mainly result of our works are:1. Put forward a common frame of AIS and its applications.2. Present a modified immune algorithm for searching multi-peaks, discuss the performance of the new algorithm according to experimental results; study effects of some key operation and parameters on performance of the algorithm.