PC Seminar: An Information-Theoretic Framework for Multiplex Graphs

Abstract: Multiplex graphs, characterised by their layered structure, exhibit informative interdependencies that are crucial for understanding complex network dynamics. Quantifying these interactions, however, remains challenging due to the non-Euclidean nature of graphs and the lack of established measures. In this talk, I will introduce an information-theoretic framework that develops multivariate measures and their consistent estimators to characterise joint complexity and capture higher-order interactions across the layers of a multiplex graph. This framework offers a systematic approach to understanding key concepts such as synergy, redundancy, and (conditional) independence, providing a theoretical perspective on complex dependencies in multiplex networks, with potential applications for pattern analysis.

Based on joint work with Sofia Olhede.

This is a seminar in our seminar series on Probability and Combinatorics (PC).