ACADEMIC LECTURE
Accurate prediction of spatiotemporal dynamical systems is vital for mitigating natural disasters, protecting health and diagnosing brain disorders early. Yet their complex mechanisms, large data volumes and high dimensionality pose challenges for simple theories.
Using the Hopf whole-brain model, we propose an adaptive method for fitting whole-brain dynamics to functional magnetic resonance imaging BOLD signals. The method distinguishes patients with major depressive disorder or autism spectrum disorder from healthy individuals more accurately than traditional approaches. We also use reservoir computing to predict solitons and complex spatiotemporal chaotic systems. Sparse measurement-based correction enables long-term prediction, and we further investigate robustness when the reservoir serves as a dynamical associative memory network.
Junjie Jiang is a distinguished researcher and doctoral supervisor at Xi’an Jiaotong University. He received a bachelor’s degree in theoretical physics from Lanzhou University in 2014 and a PhD in electrical engineering from Arizona State University in 2020. He completed postdoctoral research in computational neuroscience at New York University’s Center for Neural Science. His research interests include computational neuroscience, the intersection of nonlinear dynamics and complex systems with brain science, and olfactory brain–computer interfaces.