Error analysis for the semi-supervised algorithm under maximum correntropy criterion

Abstract As a similarity measure, correntropy has been increasingly employed in machine learning research. While numerous experimental results have shown the effectiveness of correntropy based methods, the theoretical analysis in this area is still poorly understood. In this paper, we propose a novel semi-supervised algorithm under the maximum correntropy criterion, and present an elaborate error analysis for it. An excess generalization error bound is established, which demonstrates that the proposed method is consistent, and converges at a faster rate compared with the related studies. Moreover, experiments are implemented to show the efficiency of the proposed method.