Graph Kernels: A Survey

Graph kernels have attracted a lot of attention during the last decade, and evolved into rapidly developing branch learning on structured data. During past 20 years, considerable research activity that occurred in field resulted development dozens graph kernels, each focusing specific structural properties graphs. proven successful wide range domains, ranging from social networks to bioinformatics. The goal this survey is provide unifying view literature kernels. In particular, we present comprehensive overview Furthermore, perform an experimental evaluation several those publicly available datasets, comparative study. Finally, discuss key applications outline some challenges remain be addressed.