Towards Subgraph Isomorphism Counting with Graph Kernels
Abstract: Subgraph isomorphism counting is known to be #P-complete, requiring exponential time to find an accurate solution. Recent advancements in representation learning have shown promise in representing substructures and approximating solutions. Graph kernels, which implicitly capture the correlations among substructures in diverse graphs, have demonstrated significant discriminative power in graph classification. We, therefore, explore their potential in counting subgraph isomorphisms and further enhance kernel capabilities through various variants, including polynomial and Gaussian kernels. Through comprehensive analysis, we improve the graph kernels by incorporating neighborhood information. Finally, we present the results of extensive experiments to demonstrate the effectiveness of the enhanced graph kernels and discuss promising directions for future research.
Submission Length: Regular submission (no more than 12 pages of main content)
Assigned Action Editor: ~Roman_Garnett1
Submission Number: 3125
Loading