加权线性损失孪生大间隔分布机

Weighted linear loss twin large margin distribution machine

加权线性损失孪生支持向量机(weighted linear loss twin support vector machine,WLTSVM)是针对大规模问题而构建的支持向量机(support vector machine,SVM)模型,其表现出较好的泛化性能。与SVM类似,WLTSVM中并未考虑样本集整体的间隔分布,而间隔分布已经被证明对泛化性能起着至关重要的作用。因此,为了取得更优性能,提出了一个加权线性损失孪生大间隔分布机(weighted linear loss twin large margin distribution machine,WLTLDM)。WLTLDM以一对非平行超平面模型为基础,构建了其对应的间隔均值和方差来优化间隔分布,并采用加权线性损失函数来度量类内样本的损失,使用平方损失函数来度量类间样本的损失。在真实数据集上的实验结果表明,WLTLDM是一个有效的间隔分布模型,其性能显著优于其他基准模型。

Weighted linear loss twin support vector machine(WLTSVM)is a support vector machine(SVM)model for large-scale problem,which shows good generalization performance.Similar to SVM,WLTSVM does not consider the overall margin distribution of samples set,which has been proved to play a crucial role in generalization performance.In order to achieve better performance,a weighted linear loss twin large margin distribution(WLTLDM)is proposed.Based on a pair of nonparallel hyperplane models,WLTLDM constructs the corresponding margin mean and variance to optimize the margin distribution,and uses the weighted linear loss function to measure the loss of samples within classes and the square loss function to measure the loss of samples between classes.Experimental results in real datasets show that WLTLDM is an effective margin distribution model,and its performance is significantly better than other benchmark models.