GSD: An R package for graph signal decomposition

Abstract

Graph signals residing on the vertices of a graph have recently gained prominence in research of various fields, including neural networks, social networks, traffic patterns, and sensors. Many methodologies have been proposed to analyze graph signals by adapting classical signal processing tools. In this study, we focus on graph signal decomposition, which reduces the complexity of the graph signal and increases its interpretability. Recently, several notable graph signal decomposition methods have been proposed, which include graph Fourier decomposition based on graph Fourier transform, graph empirical mode decomposition, and statistical graph empirical mode decomposition. We provide an R package GSD to efficiently implement multiscale analysis applicable to various fields. The R package GSD offers an effective tool for visualizing and decomposing graph signals.

Keywords

• Graph Fourier decomposition
• Statistical graph empirical mode decomposition
• Multiscale analysis
• R package