Network Inference

Network inference aims to reconstruct the connections within complex systems from observational data, such as time series or graph signals, with key objectives being accuracy, scalability, and robustness to noise and incomplete data. Current research emphasizes developing efficient algorithms, including those based on graph neural networks, iterative proportional fitting, and continuous-time diffusion models, to handle large-scale networks and diverse data types like single-cell data and count data. These advancements are crucial for understanding diverse systems, from biological networks and social interactions to power grids and financial markets, enabling improved prediction, control, and ultimately, a deeper understanding of complex system dynamics.