Erheng Zhong, Sun Yat-Sen University, China
Wei Fan, IBM T.J Watson Research, USA
Jing Peng, Montclair State University, USA
Olivier Verscheure, IBM T.J.Watson Research, USA
Jiangtao Ren, Sun Yat-Sen University, China

Links

Session:
Springer Link:

Abstract

The basis assumption “training and test data drawn from the same distribution” is often violated. We propose one common solution to cover various scenarios of learning under “different but related distributions” in a single framework. Examples include (a) sample selection bias (b) transfer learning and (c) uncertain training data. The main motivation is that one could ideally solve as many problems as possible with a single approach. The proposed solution extends graph transduction using the maximum margin principle over unlabeled data. The error of the proposed method is bounded even when the training and testing distributions are different. Experiment results demonstrate that the proposed method improves the traditional graph transduction by as much as 15% in accuracy and AUC in all common situations of distribution
difference. Most importantly, it outperforms, by up to 10% in accuracy, several state-of-art approaches proposed to solve specific category of distribution difference.