Unified Fourier bases for GSP on stochastic block model graphs
(2406.06306)Abstract
We consider a recently proposed approach to graph signal processing based on graphons. We show how the graphon-based approach to GSP applies to graphs sampled from a stochastic block model. We obtain a basis for the graphon Fourier transform on such samples directly from the link probability matrix and the block sizes of the model. This formulation allows us to bound the sensitivity of the Fourier transform to small changes in block sizes. We then focus on the case where the probability matrix corresponds to a (weighted) Cayley graph. If block sizes are equal, a nice Fourier basis can be derived from the underlying group. We explore how, in the case where block sizes are not equal, some or all nice properties of the group basis can be maintained. We complement the theoretical results with simulations.
We're not able to analyze this paper right now due to high demand.
Please check back later (sorry!).
Generate a summary of this paper on our Pro plan:
We ran into a problem analyzing this paper.