In this paper, we focus on how to boost the semi-supervised classification performance by exploring the multi-view graph learning. The key of multi-view graph learning is to learn a discriminative and informative graph from the multiple input graphs. However, we observe that existing multi-view graph learning methods do not sufficiently consider the diversity among views. This results in giving great weighted coefficients for mutually redundant views and affects the diversity of information of views utilized for multi-view graph learning, which finally deteriorates the semi-supervised classification performance. To address this issue, we propose a robust multi-view graph learning method with a novel and effective diversity-promoting regularized term to reduce the redundancy of views and enhance the diversity of the views. To improve the accuracy of label propagation, we further propose a unified framework which integrates multi-view graph learning, label propagation and diversity-promoting of views together. We develop an effective alternating optimization strategy to solve the optimization problem. Extensive experiments on synthetic and several benchmark data sets demonstrate the effectiveness of the proposed method.

在本文中，我们关注如何通过探索多视图图学习来提高半监督分类性能。多视图图学习的关键是从多个输入图中学习一个有判别力和信息量的图。然而，我们观察到现有的多视图图学习方法没有充分考虑视图之间的多样性。这导致为相互冗余的视图提供了很大的加权系数，并影响了用于多视图图学习的视图信息的多样性，最终降低了半监督分类性能。为了解决这个问题，我们提出了一种鲁棒的多视图图学习方法，该方法具有新颖有效的促进多样性的正则化项，以减少视图的冗余并增强视图的多样性。为了提高标签传播的准确性，我们进一步提出了一个统一的框架，将多视图图学习、标签传播和视图的多样性促进集成在一起。我们开发了一种有效的交替优化策略来解决优化问题。对合成数据集和几个基准数据集的大量实验证明了所提出方法的有效性。