多媒体网络负面信息分类方法研究与仿真

Based On Multimedia Network Negative Information Model of Optimizing SVM Classification Method Research

传统的以模式识别原理的分类机制,需要通过多次分类的方式进行负面信息的多次判断,分类次数的增加也会大幅增加错误分类的概率,导致对负面信息的分类不准。提出采用优化SVM模型算法的多媒体网络负面信息分类方法。先融合于特征选择算法给定含有两个网络负面信息特征的特征空间,使其正则化负面信息特征选择的相关性,同时采用正则化MI的平均值考量单个的网络负面信息特征和选取的负面信息特征子集冗余度,利用两个损失最小化的SVM分类器分别对训练样本进行判定,实现了多媒体网络负面信息分类。仿真结果证明,改进的方法分类精确度高,鲁棒性强。

A negative information classification method for multimedia network based on optimized SVM model algorithm is proposed. Firstly,the feature selection algorithm is integrated to give the feature space,which contains two negative information features of the network,so that the correlation of feature selection of the negative information is made regularization. At the same time,the average value of regularization MI is used to consider the single negative information feature of network,and select the redundant degree of negative information feature subset. The training samples are determined respectively by using two SVM classifiers with minimized losses,and the negative information classification of the multimedia network is realized. The simulation results show that the improved method has high classification accuracy and strong robustness.