针对离散二进制粒子群(binary particle swarm optimization,BPSO)算法在解决SVM集成选择问题时容易早熟的问题,提出了一种文化算法架构下的多种群协作算法(Ca-MultiPop)。结合BPSO算法的快速演化能力,利用遗传算法(genetic algorithm,GA)增加种群的多样性;在两种进化算法中使用不同的适应度函数,兼顾了集成精度和基分类器之间的差异性。仿真结果表明,该算法在计算精度方面相对于BPSO算法在解决SVM集成选择问题时有所提高。
For the premature problem of BPSO algorithm in solving SVM selection ensemble,a multiple population collaboration algorithm in the framework of culture algorithm is proposed.BPSO's premature is avoided by the hybrid of BPSO's quick evolvement and GA's diversity of populations.Meanwhile,different fitness functions are used in BPSO and GA to take account of the difference between ensemble precision and base classifiers.Simulation results show that the algorithm proposed is superior to BPSO algorithm in precision and efficiency when solving SVM selection ensemble problem.