Multi-view Laplacian sparse feature selection for web image annotation

Abstract

Recently, semi-supervised sparse feature selection, which can exploit the large number unlabeled data and small number labeled data simultaneously, has placed an important role in web image annotation. However, most of the semi-supervised feature selection methods are developed for single-view data, which can not reveal and leverage the correlated and complemental information between different views. Recently, multi-view learning has obtained much research attention, so we apply multi-view learning into semi-supervised sparse feature selection and propose a multi-view semi-supervised sparse feature selection method based on graph Laplacian, namely Multi-view Laplacian Sparse Feature Selection (MLSFS) in this paper. MLSFS can realize sparse feature selection by utilizing the correlated and complemental information between different views. A simple iterative method is proposed to solve the objective function of MLSFS. We apply our algorithm into image annotation and conduct experiments on two web image datasets. The results show that the proposed multi-view method outperforms the single-view methods.