Deep Graph Kernels

A unified framework to learn latent representations of sub-structures for graphs, inspired by latest advancements in language modeling and deep learning, which achieves significant improvements in classification accuracy over state-of-the-art graph kernels.