基于非均匀变异算子的状态空间进化算法-《计算机技术与发展》

文章信息/Info

Title:: State Space Evolutionary Algorithm Based on Non-uniform Mutation Operator

Author(s):: LING Zhe; LI Mao-jun; School of Electrical and Information Engineering,Changsha University of Science &Technology,Changsha 410114,China

Keywords:: state space algorithm; transfer matrix; value of fitness; big data

摘要:: 基于非均匀变异算子的状态空间进化算法(NUMSEA)是一种具有新颖性的实数编码进化算法。针对传统的状态空间进化算法转移矩阵的不足,设计一种基于非均匀变异等算子改进的状态空间转移矩阵。该矩阵突破了传统的状态空间转移矩阵,并在此基础上增加了非均匀变异算子以及非均匀算术交叉算子。通过提取分析每一代的最适值,再左乘新的转移矩阵,能够在原有的最优个体附件进行微小的搜索。进一步实现了转移矩阵随群体中个体适应度值的自适应变化,上一代群体中适值越大的个体在生成新个体时所作的贡献越大,算法的收敛速度也将增加。实验结果表明,改进算法不仅能提升对主效基因挖掘的精确性与平稳性,还能缩短对特征数据的提取时间。

Abstract:: State space evolutionary algorithm based on non-uniform mutation (NUMSEA) is a novel evolutionary algorithm with real number coding. Aiming at the disadvantage of the transfer matrix of traditional state space model,we design an improved state space transfer matrix based on non-uniform mutation,which breaks through the traditional state space transfer matrix and on the basis adds nonuniform mutation operator and non-uniform arithmetic crossover operator. By extracting and analyzing the optimum value of each generation,left multiplying by newly transfer matrix,we can conduct a small search in the original optimal individual attachment. The adaptive change of transfer matrix is further implemented with the individual fitness value in groups. In the previous generation,the more adaptable individuals in the group,the more contributions they make in generating new individuals,and the convergence speed of the algorithm will also increase. The experiment shows that the improved algorithm can not only improve the accuracy and stability of the majorgene mining,and shorten the time of extracting characteristic data.