基于粒子群算法的投影孪生支持向量机

Projection Twin Support Vector Machine Based on Particle Swarm Optimization

投影孪生支持向量机(Projection Twin Support Vector Machine,PTSVM)是一种有效的分类算法,存在参数选择困难和算法总体运行时间过长等问题.文章将粒子群算法(Particle Swarm Optimization,PSO)和PTSVM相结合,提出一种基于粒子群算法的投影孪生支持向量机(PSO-PTSVM).该算法能够快速寻找最优参数,较好地解决PTSVM参数选择困难的问题.在UCI数据集上验证PSO-PTSVM的有效性,实验结果表明,PSO-PTSVM运行总时间短,在大多数数据集上具有较高的分类精度.

Projection twin support vector machine(PTSVM)is an effective classification algorithm,but it has some problems such as the difficulty of parameter selection and the long running time.In this paper,particle swarm optimization and PTSVM are combined,and a projection twin support vector machine based on parti⁃cle swarm optimization is proposed.The algorithm can quickly find the optimal parameters and overcome the problem of parameter selection of PTSVM.The validity of PSO-PTSVM is verified on UCI data sets.The exper⁃imental results show that PSO-PTSVM has shorter running time and higher classification accuracy.