Work

Global graph kernels using geometric embeddings

Year: 2014
Type: article
Abstract: Applications of machine learning methods in-creasingly deal with graph structured data through kernels. Most existing graph kernels compare graphs in terms of features defined on small subgraphs such as walks, paths or graphlets, adopting an inherently local perspec-tive. However, several interesting properties such as girth or chromatic number are global proper-ties of the graph, and are not captured in local substructures. This paper presents two graph ker-nels defined on unlabeled graphs which capture global properties of graphs using the celebrated Lovász number and its associated orthonormal representation. We make progress towards the-oretical results aiding kernel choice, proving a result about the separation margin of our kernel for classes of graphs. We give empirical results on classification of synthesized graphs with im-portant global properties as well as established benchmark graph datasets, showing that the ac-curacy of our kernels is better than or competitive to existing graph kernels. 1.
Language: English
Open Access status: green
OpenAlex ID: W2107726887
FWCI: 3.505
Cites: 44
Cited by: 76
Related to: 10