Automated Modelling of Complex Systems

We build models for complex systems according to the time series of the observable behaviors of the system in an automatic way. By using techniques like Graph (Neural) Network, Causal Inference, Information Theory, and Deep Learning etc., to learn the dynamics, infer the un-observable network structures and node states, and identify the phenomenon of emergence.


Machine Learning on Complex Networks

Machine learning provides us with new tools to handle and solve hard problems on complex networks, like identifying node types, classifying complex networks, and completing a partial observable network.


Open Flow Networks

By modeling a complex open system as a directed weighted network and adding two special nodes, source and sink, into the network, we can study the distributions of flows within a system and model the general flows in wide complex open systems, including the attention flows of online education and learning, online forum, and web sites; product flows of international trading system, and energy flows on ecological systems, etc.


Scaling Analysis

In urban systems, biology systems, firms, and web forums, by discovering scaling laws on the aggregated level from the data, scaling analysis can characterize the macroscopic universal patterns of complex systems and provide some insights for the micro-level mechanisms.



Research Group


School of Systems Science
Beijing Normal University, 100875
Beijing, China