A perturbation-based approach to identifying potentially superfluous network constituents

TL;DR

This paper introduces a perturbation-based method utilizing vertex and edge centrality to identify potentially superfluous constituents in networks derived from empirical time series data, aiming to improve network interpretation.

Contribution

The paper presents a novel perturbation-based approach for detecting superfluous network constituents, addressing issues caused by oversampling in network construction from time series.

Findings

Effective identification of superfluous constituents in various network types

Method performs well on small-world, scale-free, random, and complete networks

Improves accuracy of network analysis by filtering unnecessary elements

Abstract

Constructing networks from empirical time series data is often faced with the as yet unsolved issue of how to avoid potentially superfluous network constituents. Such constituents can result, e.g., from spatial and temporal oversampling of the system's dynamics, and neglecting them can lead to severe misinterpretations of network characteristics ranging from global to local scale. We derive a perturbation-based method to identify potentially superfluous network constituents that makes use of vertex and edge centrality concepts. We investigate the suitability of our approach through analyses of weighted small-world, scale-free, random, and complete networks.