Title: Total stability of support vector machines
Authors: Daohong Xiang - Zhejiang Normal University (China) [presenting]
Andreas Christmann - University of Bayreuth (Germany)
Ding-Xuan Zhou - City University of Hong Kong (Hong Kong)
Abstract: Some total stability results for SVMs are established, which show that SVMs based on kernels are even stable, if the full triple (P, $\lambda$, k) consisting of the underlying probability measure P, the regularization parameter $\lambda,$ and the kernel k changes slightly.