Machine Learning for Predicting the Dynamics of Complex Networks
Abstract
In recent years, machine learning methods such as “deep learning” have proven enormously successful for tasks such as image classification, voice recognition, and more. Despite their effectiveness for big-data classification problems, these methods have had less success for time series prediction, especially for complex systems like those we see in brain dynamics or social dynamics. In this talk, I will discuss how a Reservoir Computer (RC) – a special kind of artificial neural network that offers a “universal” dynamical system – can draw on its own internal complex dynamics in order to forecast complex systems. Like many other machine learning architectures, RCs provide a knowledge-free approach because they forecast purely from past measurements without any specific knowledge of the system dynamics. By building a scheme that judiciously combines the knowledge-free prediction of the RC with a knowledge-based model, we demonstrate a dramatic improvement in forecasting complex systems. We show that we can forecast the dynamics of not only individual time series but of large complex networks by constructing a parallel RC architecture that mimics the topology of the network of interest. We achieve good predictions in cases where the network links are known and also in cases where the network links are unknown and inferred via a data-driven approach to approximately optimize prediction.
About the Speaker
Michelle Girvan is a Professor in the Department of Physics at the University of Maryland (UMD). She also has appointments in the Institute for Physical Science and Technology, the Institute for Research in Electronics and Applied Physics, and the Applied Math and Scientific Computing Program. Her research focuses on applications of network science to biological, social, and technological systems. Much of her recent work is aimed at the intersection of network science and machine learning. Girvan serves as Director and PI of the COMBINE graduate program in Network Biology. Girvan received bachelor’s degrees in physics and mathematics from MIT. She then went on to earn her PhD in physics from Cornell University followed by a postdoctoral fellowship at the Santa Fe Institute before joining the faculty at the University of Maryland. Girvan serves on the Santa Fe Institute’s External Faculty and Science Steering Committee. Girvan is a Fellow of the American Physical Society and the Network Science Society. In 2022, she was named a Distinguished ScholarTeacher by the University of Maryland.