A0614
Title: On learning with Gaussian kernel
Authors: Fusheng Lv - Nankai University (China) [presenting]
Abstract: Gaussian kernel-based learning has received great success in practice over the last decades. The theory properties of two types of learning algorithms is discussed generated by the Tikhonov regularization scheme and associated with varying Gaussian kernels. It shows that the algorithm with corrector loss achieves the minimax rate of convergence. The convergence rate of regression risk for the modal regression algorithm will also be presented.
