自适应双正则化支持向量机的解路算法
Solution Path Algorithm for the Doubly Regularized Support Vector Machine
-
摘要: 针对自适应双正则化支持向量机,证明了其最优解关于单正则化参数是分段线性的,并据此提出了完全正则化解路算法,最后,通过在急性白血病数据集上进行分类实验,验证了所提算法的有效性Abstract: For the adaptive doubly regularized support vector machine, it was proved that its solution path is piecewise linear with respect to one regularized parameter when another parameter is fixed. Based on this result, an entire regularized solving path algorithm was proposed. The experiment results on the binary classification of acute leukemia demonstrated the effectiveness of the proposed algorithm.