Fast Graph Kernel with Optical Random Features

Hashem Ghanem (1) , Nicolas Keriven (1, 2) , Nicolas Tremblay (1)

1 GIPSA-GAIA - GIPSA Pôle Géométrie, Apprentissage, Information et Algorithmes
2 CNRS - Centre National de la Recherche Scientifique

The graphlet kernel is a classical method in graph classification. It however suffers from a high computation cost due to the isomorphism test it includes. As a generic proxy, and in general at the cost of losing some information, this test can be efficiently replaced by a user-defined mapping that computes various graph characteristics. In this paper, we propose to leverage kernel random features within the graphlet framework, and establish a theoretical link with a mean kernel metric. If this method can still be prohibitively costly for usual random features, we then incorporate optical random features that can be computed in constant time. Experiments show that the resulting algorithm is orders of magnitude faster that the graphlet kernel for the same, or better, accuracy.

Mots clés: Optical random features; Graph kernels

Domaines: Apprentissage [cs.LG]

Hashem Ghanem, Nicolas Keriven, Nicolas Tremblay. Fast Graph Kernel with Optical Random Features. ICASSP 2021 - IEEE International Conference on Acoustics, Speech, and Signal Processing, Jun 2021, Toronto, Canada. ⟨10.1109/ICASSP39728.2021.9413614⟩. ⟨hal-02976716⟩