Linkage beforehand procesing and genetic stream

The linkage learning genetic algorithm (LLGA) and equilibrium genetic algorithm (EGA) have improved efficiency. LLGA emphasizes the importance of linkage in the process of evolution. However, because of the random selection of initial population, evolutionary speed may be unnecessarily slowed down. Besides, linkage between genes may become closer and closer because of the pressure of selection, which makes the harmful genes become closely connected with the favorable ones and reduces the possibility of being erased. By reserving the optimal chromosome of each generation, EGA improves the converging speed. However, the speed of mutation is far slower than that of selection, resulting in premature convergence. In light of theory and methods of genetics and evolutionary biology, this paper proposes the idea of utilizing linkage processing to initialize the population to improve the efficiency of the genetic algorithm. The genetic stream is proposed to replace the mutation operator to keep the diversity of population and avoid premature convergence. We call this method an advanced linkage learning genetic algorithm (ALLGA). Empirical comparisons between LLGA and ALLGA are performed to testify the effectiveness and accuracy of ALLGA.