Accepted Paper: Joint Consensus and Diversity for Multi-view Semi-supervised Classification

Back to list of accepted papers


Wenzhang Zhuge (National Universify of Defense Technology); Chenping Hou (National University of Defense Technology); Shaoliang Peng (Hunan University); Dongyun Yi ()


With the advance of data acquisition ways, there are increasingly growing number of multi-view data. Considering the high price of labeling data in many machine learning applications, we focus on multi-view semi-supervised classification problem. To address this problem, in this paper, we propose a method called Joint Consensus and Diversity for Multi-view Semi-supervised Classification (JCD), which learns a common label matrix for all training samples and view-specific classifiers simultaneously. A novel classification loss named probabilistic square hinge loss is proposed, which avoids the incorrect penalization problem and characterizes the contribution of training samples according to its uncertainty. Power mean is introduced to incorporate the losses of different views, which contains the auto-weighted strategy as a special case and distinguishes the importance of various views. To solve the non-convex minimization problem, we prove that its solution can be obtained from another problem with introduced variables. And an efficient algorithm with proved convergence is developed for optimization. Extensive experimental results on six datasets demonstrate the effectiveness of the proposed algorithm.