最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)通过求解一个线性等式方程组来提高支持向量机(Support Vector Machine,SVM)的运算速度.但是,LSSVM没有考虑间隔分布对于LSSVM模型的影响,导致其精度较低.为了增强LSSVM模型的泛化性能,提高其分类能力,提出一种具有间隔分布优化的最小二乘支持向量机(LSSVM with margin distribution optimization,MLSSVM).首先,重新定义间隔均值和间隔方差,深入挖掘数据的间隔分布信息,增强模型的泛化性能;其次,引入权重线性损失,进一步优化了间隔均值,提升模型的分类精度;然后,分析目标函数,剔除冗余项,进一步优化间隔方差;最后,保留LSSVM的求解机制,保障模型的计算效率.实验表明,新提出的分类模型具有良好的泛化性能和运行时间.
Least square support vector machine with margin distribution optimization
Least squares support vector machine(LSSVM)improves the speed of the support vector machine(SVM)by solving a system of linear equations.However,LSSVM does not consider the effect of margin distribution on the LSSVM model,resulting in low accuracy.In order to enhance the generalization performance of the LSSVM and improve its classification ability,a LSSVM with margin distribution optimization(MLSSVM)is proposed.Firstly,the margin mean and margin variance are redefined to deeply mine the distribution information of the data,enhancing the generalization performance of the new model.Secondly,weighted linear loss is introduced,which can further optimize the margin mean,improving the classification accuracy of the new model.Then,the objective function is analyzed to remove the redundant terms,making the margin variance be further optimized.Finally,the solving mechanism of the LSSVM is retained to ensure the computational efficiency of the new model.Experimental results show that the new proposed classification model has good generalization performance,classification accuracy,and running time.
least squares support vector machinelarge margin distribution machinemargin distribution optimizationweighted linear loss
万方数据
维普
