Identifying Core-Periphery Structures Using Graph Neural Networks

Kumar, R and Gurugubelli, S and Chepuri, SP (2022) Identifying Core-Periphery Structures Using Graph Neural Networks. In: 56th Asilomar Conference on Signals, Systems and Computers, ACSSC 2022, 31 October- 2 November 2022, Virtual, Online, pp. 251-255.

PDF acssc_2022.pdf - Published Version Restricted to Registered users only Download (824kB) | Request a copy

Abstract

In this paper, we develop a graph neural network (GNN) model for identifying core-periphery structures in graphs. Existing core score estimation algorithms to identify core nodes in graphs are optimization-based models and ignore any complementary information that the node attributes might carry about the coreness of nodes or cannot handle the cases when the graph is only partially observed. We propose a GNN model to identify the core and periphery nodes in a graph and simultaneously impute any missing edges while preserving the core-periphery structure in the graph. Numerical experiments on various synthetic and real data show that the proposed method successfully identifies the core and peripheral nodes from a partially observed graph while imputing edges such that the core-periphery structure in the graph is intact, in contrast to the standard GNN-based link prediction algorithms, which fail to preserve the structure.

Item Type:	Conference Paper
Publication:	Conference Record - Asilomar Conference on Signals, Systems and Computers
Publisher:	IEEE Computer Society
Additional Information:	The copyright for this article belongs to IEEE.
Keywords:	Computer vision; Graph theory; Neural network models; Numerical methods, Core nodes; Core peripheries; Core score; Core-periphery network; Estimation algorithm; Graph neural networks; Link prediction; Neural network model; Optimisations; Structured graphs, Graph neural networks
Department/Centre:	Division of Electrical Sciences > Electrical Communication Engineering
Date Deposited:	24 Apr 2023 10:27
Last Modified:	24 Apr 2023 10:27
URI:	https://eprints.iisc.ac.in/id/eprint/81285

Actions (login required)

View Item