论文部分内容阅读
针对音频信号准确性分类的问题,提出一种基于改进的的粒子群优化算法(PSO)的支持向量机(SVM)音频信号分类的方法,简称IPSO-SVM.首先用Mel倒谱系数法对4种音频信号进行特征提取.其次在PSO中引入自适应变异因子,能够成功地跳出局部极小值点;然后对PSO中的惯性权重进行了改进,将惯性权重由常数变为指数型递减函数.随着迭代的进行,使权重逐渐减小,这样做有利于粒子进行局部寻优.最后用改进的PSO不断优化SVM中的惩罚因子c和核函数参数g来提高预测精度.实验结果表明,与传统的SVM、PSO-SVM、GA-SVM相比,我们提出的IPSO-SVM算法分类结果更精确.
In order to solve the problem of classifying the accuracy of audio signals, this paper proposes a method based on improved Particle Swarm Optimization (PSO) support vector machine (SVM) audio signal classification, referred to as IPSO-SVM. Then the audio signal is extracted by feature extraction.Secondly, adaptive mutation factor is introduced into PSO to jump out of local minima. Then the inertia weight in PSO is improved and the inertia weight is changed from constant to exponential decreasing function. As the iteration proceeds, the weight is gradually reduced, which is good for the local optimization of the particle.Finally, the improved PSO is used to optimize the penalty factor c and the kernel function parameter g in SVM to improve the prediction accuracy.The experimental results show that, Compared with traditional SVM, PSO-SVM and GA-SVM, the classification results of IPSO-SVM proposed by us are more accurate.