Fast Multiscale Diffusion on Graphs
Signal Processing
2021-05-03 v1 Numerical Analysis
Numerical Analysis
Abstract
Diffusing a graph signal at multiple scales requires computing the action of the exponential of several multiples of the Laplacian matrix. We tighten a bound on the approximation error of truncated Chebyshev polynomial approximations of the exponential, hence significantly improving a priori estimates of the polynomial order for a prescribed error. We further exploit properties of these approximations to factorize the computation of the action of the diffusion operator over multiple scales, thus reducing drastically its computational cost.
Cite
@article{arxiv.2104.14652,
title = {Fast Multiscale Diffusion on Graphs},
author = {Sibylle Marcotte and Amélie Barbe and Rémi Gribonval and Titouan Vayer and Marc Sebban and Pierre Borgnat and Paulo Gonçalves},
journal= {arXiv preprint arXiv:2104.14652},
year = {2021}
}