Inference and optimal control of network dynamics

Building on the dynamic message-passing (DMP) approach, I develop computationally efficient algorithms to predict and control spreading processes on complex networks.

Interacting spreading processes

I developed a DMP-based algorithm to predict the dynamics of multiple interacting (collaborative or competitive) spreading processes on sparse networks. The framework captures how coupled contagions reshape each other’s dynamics beyond what standard independent models can describe.

Optimal control under budget constraints

I further proposed an optimisation framework for the control of interacting spreading processes via resource allocation under realistic budget constraints and finite time horizons. The resulting algorithms can both enhance desirable diffusion (e.g., information or awareness campaigns) and suppress harmful spread (e.g., coupled epidemic strains), outperforming standard topology-based strategies for epidemic control.

Related publications

• Competition, Collaboration, and Optimization in Multiple Interacting Spreading Processes — Physical Review X