National Natural Science Foundation of China(No.60705004)
By utilizing hyperbolic tangent function,we constructed a novel hyperbolic tangent loss function to reduce the influences of outliers on support vector machine(SVM)classification problem.The new loss function not only...
针对一类分类马氏椭球学习机(Mahalanobis ellipsoidal learning machine for one class classification,MELM)方法中选取参数C比较困难的问题,提出一种改进的方法-MELM。这种方法通过引入一个具有明确物理意义的参数,即是超椭球外部的...