A New Self-adjusting Immune Genetic Algorithm

The genetic algorithm based on immunity has recently been an appealing research methodology in evolutionary computation. Aiming to cope with the problems of genetic algorithms, i.e., the solution is apt to trap into a local optimum and the convergence speed is slow, this paper proposes a new self-adjusting immune genetic algorithm, called SaiGa ( S elf- a djusted i mmune G enetic a lgorithm), which seeks for an optimal solution with regard to complex problems such as the optimization of multidimensional functions by automatically tuning the crossover and the mutation probabilities, which can help avoid prematurity phenomena and maintain individual diversity. In particular, SaiGa introduces a variable optimization approach to improve the precision in terms of solving complex problems. The empirical results demonstrate that SaiGa can greatly accelerate convergence for finding an optimal solution compared with genetic algorithms and immune algorithms, achieve a better precision in function optimization, and avoid prematurity convergence.