An Iterative Approach to Data-Driven Inference for Decoding Oscillator Network Structures

Abstract

In complex networks, interactions between multiple agents give rise to an array of intricate global dynamics, ranging from synchronization to cluster formations. Decoding the connectivity structure as well as the types of interactions from measurement data is the first step toward understanding these intriguing behaviors. In this paper, we present a bilinear optimization framework to infer both the connectivity and interaction functions of oscillator networks with the identical class of coupling functions. We then propose an iterative algorithm to solve the resulting bilinear problem and illustrate its convergence. We validate our approach on both simulated and noisy experimental datasets, where we demonstrate its effectiveness compared with existing approaches.