基于PSO优化的SMO算法研究及应用

顺序优化(SMO)是支持向量机(SVM)的一种有效训练算法,但SMO的参数选择问题是算法性能优劣的关键所在,只有选择了合适的参数才能使算法性能达到最优。因此,在详细介绍了SMO算法的基础上,着重研究了基于微粒群优化(PSO)的SMO算法的双层优化原理,并通过仿真进行了应用研究,将该方法的有效性进行了验证。实验结果表明,经过PSO优化的SMO算法与其他算法相比具有更高的准确性。

Research and application of SMO algorithm based on PSO optimization

Sequential optimization(SMO) is an effective training algorithm of SVM,but the selection of the SMO algorithm parameters is the key to the performance,and only choosing the appropriate algorithm parameters can achieve optimal performance.Therefore,on the basis of describing the SMO algorithm in detail,we have studied the bilevel optimization theory which is about the SMO algorithm based on PSO and testified the validities of this method by simulation studies.The results of the experiments show that compared with other algorithms,the SMO algorithm optimized by PSO has a higher accuracy.