Combinative hypergraph learning for semi-supervised image classification

Recent years have witnessed a surge of interest in hypergraph-based transductive image classification. Hypergraph-based transductive learning models the high-order relationship of samples by using a hyperedge to link multiple samples. In order to extend the high-order relationship of samples, we incorporate linear correlation of sparse representation to hypergraph learning framework to improve learning performance. In this paper, we present a new transductive learning method called combinative hypergraph learning (CHL). CHL captures the similarity between two samples in the same category by adding sparse hypergraph learning to conventional hypergraph learning. And more, we propose two strategies to combine the two hypergraph learning methods. Experimental results on two image datasets have demonstrated the effectiveness of CHL in comparison to the state-of-the-art methods and shown that our proposed method is promising.