Improving (1+λ)-ES Using the Clonal Selection Theory of Immunity

摘要

In the evolvable hardware,apart from the influence of evolutionary model and the chromosome encoding,the evolutionary algorithm directly influences the evolution speed.The core idea of the traditional (1+λ) algorithm is very appropriate for the evolvable hardware.However,after the evolution on a mass of different target circuits,we find a serious problem of random number seed relying on the (1+λ) algorithm.In other words,there is a wide gap between the evolution generations of different random number seeds.This Paper improves the (1+λ) algorithm combining the clonal selection theory of immunity and experimental data analysis.A large number of experiments demonstrate that the improved method dramatically improves the speed of evolution and made the distribution of the evolution generation more concentrated compared with the traditional algorithm.