混沌粒子群算法对支持向量机模型参数的优化

Chaos Particle Swarm Optimization Algorithm for Optimizing the Parameter of SVM

研究支持向量机模型优化问题,支持向量机的参数选择决定了其学习性能和泛化能力,由于在参数的选择范围内可选择的数量很多,在多个参数中进和盲目搜索最优参数是需要极大的时间代价,并且很难得到最优参数。常用的支持向量机优化方法有遗传算法、粒子群算法都存在易陷入局部极值,优化效果较差。为解决支持向量机参数寻优问题,提出一种基于混沌粒子群的支持向量机参数选择方法。将混沌理论引入粒子群优化算法中,从而提高种群的多样性和粒子搜索的遍历性,从而有效地提高了PSO算法的收敛速度和精度,得了优化支持向量机模型。并以信用卡案例数据作为研究对象进行了仿真,实验结果表明,混沌粒子群优化的SVM分类器比传统算法优化的SVM分类器的精度高和更高的效率,应用效果好。

The parameters of support vecto rmachine decide its study performance and generalizing ability.As the parameter choice is infinite,the parameter chioce needs enormous time,and is very difficult to approach superiorly.At present,genetic algorithm and particle swarm optimization algorithm are common optimization algorithm for SVM.However,these methods are easy to lapse into local extremum,so that optimization result might be bad.Thus,in the study,chaos thought is introduced in particle swarm optimization algorithm,and chaos particle swarm optimization is presented to optimize support vecto rmachine.Chaos particle swarm optimization can improve diversity of swarm and ergodic property of particle,which improves convergent speed and accuracy of particle swarm optimization,and can optimize SVM well.And the credit card company data are used as research object to the classification performance of chaos particle swarm optimization-based support vecto rmachine.The experimental results show that support vector machine optimized by chaos particle swarm optimization has higher classification accuracy than optimized by particle swarm optimization.