Abstract

Dynamical systems describe the evolution of complex phenomena, from neuron firing and turbulence to markets and materials. Yet their nonlinear and often chaotic nature makes long-term prediction and discovery a major challenge. In this talk, I will show how artificial intelligence can serve as a powerful tool for dynamical systems research—extending prediction horizons in chaotic flows, uncovering hidden conservation laws, and modeling systems with abrupt transitions. At the same time, by viewing neural networks as dynamical systems, we identify hyperbolic chaos as the source of their instability, and introduce control strategies that suppress it. Together, these results highlight a two-way connection: using AI to advance dynamics, and dynamics to build more stable and interpretable AI.

Biography

Prof. Hongkun Zhang is Chair Professor at the Greater Bay Area University. She works at the intersection of dynamical systems and artificial intelligence, with contributions spanning both the mathematical theory of chaos and stability in neural networks (Dynamics for AI) and the use of AI to discover new structures in complex systems (AI for Dynamics). She previously held a tenured professorship at the University of Massachusetts Amherst and has received honors including the NSF CAREER Award and the Simons Fellowship.