【摘要】 给出了一种利用目标函数的二阶信息选择工作集训练加权支持向量机的算法,导出了加权支持向量机的KKT条件。实验结果表明,与利用目标函数的一阶近似信息选择工作集的训练算法相比,该算法减少了训练迭代次数,特别是训练集规模较大时,该算法的收敛速度有较大幅度的提高。更多还原

【Abstract】 This paper proposed a new traning algorithm of W-SVM that uses second-order information of objective function to select working set and deduced its KKT optimization condition.The experimental results show that the algorithm reduces the number of iterations,comparing with the training algorithm that uses first-order approximate information of objective function to select working set,especially when the training set is very large,its convergence rate is speeded up highly.更多还原