Geodesics of Dynamic Graphs for Regime Change Detection
Researchers have developed a new method for detecting regime changes in dynamic graphs, moving beyond traditional abrupt transition models. This approach defines regimes as periods of coherent dynamics, characterized by trajectories along geodesics in a graph space. By measuring drifts in these dynamics, the method can identify significant changes, outperforming existing models on various experiments and aligning detected changes with external events like the COVID-19 pandemic. AI
IMPACT Introduces a novel framework for analyzing evolving network dynamics, potentially improving real-world applications in social networks and mobility analysis.