Sub-network Based Kernels for Brain Network Classification

Biao Jie, Mingxia Liu, Xi Jiang, Daoqiang Zhang

Abstract: In brain network analysis, a challenging problem is deciding how to measure the similarity between a pair of networks. Recently, graph kernels have been proposed for measuring the similarity between brain networks. However, existing graph kernels are mainly defined on general graphs that ignores specific characteristics of brain networks, such as the uniqueness of nodes (i.e., each node corresponds to a unique brain region). Accordingly, in this paper, we construct a novel sub-network based kernel for brain networks and apply it for mild cognitive impairment (MCI) classification. Experimental results on a real MCI dataset demonstrate that the proposed method outperform several state-of-the-art graph kernel based methods.