Data collected from generic complex systems can be represented in the form of networks, ranging from online social networks, physical contact networks to critical infrastructures. My team focuses on Network Data Science, which aims to develop methodologies to predict, model and control processes (such as information, disease, failure propagation and social or financial contagion) on networks combining data and network science approaches.
The prediction problems addressed include the prediction of late payment of invoices, default and KPI measures of companies in the network of monetary transactions among companies and the prediction of outbreak size of information/epidemic spreading. Our expertise in spreading process has been utilized to improve the design of temporal network embedding algorithm for general link prediction and classification tasks.
Modeling and Control: We have developed methodologies, both theoretical and empirical, to understand how the underlying interdependent and time-evolving network topology influence a dynamic process. Such understanding enables the optimization of network topology to be robust against virus and failure propagation or efficient in information diffusion. Furthermore, strategies to mitigate the spread of epidemics or opinion via e.g. reducing time-evolving network interactions and information spread have be developed.
Our ambition is to discover the underlying mechanism or process of a complex social-physical system that we don’t understand via network data science. Such interpretation of data to the extent that we could further optimize the system is deemed as the fore-runner of AI-Networking. This ambition is being pursued via her current projects like NOW-TOP and KPN-TUDelft AI Networking project.