Abstract

Hypernetworks, which enable joint interactions among three or more nodes, reveal complex network dynamics underlying various systems. This study focuses on reconstructing hypernetwork connectivity structures from time-series data. By transforming the inference task into a bilinear optimization problem, we develop an iterative algorithm to recover adjacency matrices and coupling strengths accurately. Our method's robustness is validated with phase-coupled oscillator models, achieving an average AUC score of 0.8 for connectivity reconstruction. Future directions include enhancing algorithm efficiency for larger networks and refining coupling strength estimations. This work advances network inference by offering a scalable and accurate approach to decoding higher-order network interactions.

Document Type

Article

Author's School

McKelvey School of Engineering

Author's Department

Electrical and Systems Engineering

Class Name

Electrical and Systems Engineering Undergraduate Research

Language

English (en)

Date of Submission

12-12-2024

Share

COinS