A Generative Model with Ensemble Manifold Regularization for Multi-view Clustering

Abstract

Topic modeling is a powerful tool for discovering the underlying or hidden structure in documents and images. Typical algorithms for topic modeling include probabilistic latent semantic analysis (PLSA) and latent Dirichlet allocation (LDA). More recent topic model approach, multi-view learning via probabilistic latent semantic analysis (MVPLSA), is designed for multi-view learning. These approaches are instances of generative model, whereas the manifold structure of the data is ignored, which is generally informative for nonlinear dimensionality reduction mapping. In this paper, we propose a novel generative model with ensemble manifold regularization for multi-view learning which considers both generative and manifold structure of the data. Experimental results on real-world multi-view data sets demonstrate the effectiveness of our approach.