摘要
This paper is a continuation of the study of classification learning algorithms generated by regularization schemes associated with Gaussian kernels and general convex loss functions. In previous papers Xiang and Zhou (2009) [5], Xiang (2010)171, it is assumed that the convex loss phi has a zero. This excludes some useful loss functions without zero such as the logistic loss l(t) = log(1 + exp(-t)). The main purpose of this paper is to conduct error analysis for the classification learning algorithms associated with such loss functions. The learning rates are derived by a novel application of projection operators to overcome the technical difficulty.
学科与关键词
关键词
Reproducing kernel Hilbert spaceBinary classificationLogistic lossSobolev spaceProjection operator
- Classification with Gaussians and Convex LossXiang, Dao-Hong*; Zhou, Ding-XuanJOURNAL OF MACHINE LEARNING RESEARCH, 2009, 10: 1447-1468.被引 36
- Online Classification with Varying GaussiansHu, Ting; Zhou, Ding-Xuan*Studies in Applied Mathematics, 2010, 124(1): 65-83.被引 2
- Statistical Rates of Convergence for Functional Partially Linear Support Vector Machines for ClassificationZhang, Yingying; Zhao, Yan-Yong; Lian, Heng*JOURNAL OF MACHINE LEARNING RESEARCH, 2022, 23.
