免疫粒子群算法下向量机参数选择及金融应用

Parameter Selection of Support Vector Machines Based On Immune Particle Swarm Optimization Algorithm and Its Application in Finance

针对粒子群算法易陷入局部最优值的缺点,将免疫原理引入粒子群算法中,利用免疫记忆与自我调节机制促使各适应度层次的粒子维持一定浓度,保证群体的多样性,从而避免算法陷入局部最优。随后将这种改进的算法应用于支持向量机参数的选择,并在Breast Cancer等数据集上进行了实验,实验结果表明利用免疫粒子群算法选取支持向量机最优参数,能够提高支持向量机的分类正确率,具有一定的实用性,特别在经济金融应用上前景可观。

To avoid trapping into local optimization of Particle Swarm Optimization （PSO） algorithm, the principle of immune was introduced to improve the PSO algorithm for searching the optimal parameters of support vector machines （SVM）.The improved method utilized the function of immune memory and the self adjustment mechanism to maintain the concentration of particles at a certain level in every layer to guarantee the diversity of population. So it avoided the problem of local optimization. The improved algorithm was verified with the Breast Cancer, Ionosphere and German datasets. The results demonstrate that the algorithm can improve the overall performance of SVM classifier and its application in the field of finance will lead to prosperous future.