Immune Genetic Algorithm Optimization and Application

This paper discusses the limitations of the genetic algorithm and the unique advantages of immune genetic algorithm. A clustering based vaccine extraction algorithm is proposed which is being proved to be efficient and reasonable. The clustering based selection method is used to reduce the similarity between antibodies and avoid falling into local optimal solution. In order to speed up the convergence rate, elite antibodies are trained by the steepest descent method. Finally, the new algorithm is applied to the neural networks based Chinese word segmentation model. Experiments show that the accuracy achieved by the new algorithm is much higher than the traditional BP algorithm.