Deep kernel supervised hashing for node classification in structural networks

Node classification in structural networks has been proven to be useful many real world applications. With the development of network embedding, performance node greatly improved. However, nearly all existing embedding based methods are hard capture actual category features a because linearly inseparable problem low-dimensional space; meanwhile they cannot incorporate simultaneously structure information and label into embedding. To address above problems, this paper, we propose novel Deep Kernel Supervised Hashing (DKSH) method learn hashing representations nodes for classification. Specifically, deep multiple kernel learning is first proposed map suitable Hilbert space deal with problem. Then, instead only considering similarity between two nodes, matrix designed merge both information. by matrix, learned preserve kinds well from space. Extensive experiments show that significantly outperforms state-of-the-art baselines over three benchmark datasets.