معرفی
Yuqing Wang is a postdoctoral researcher in the Department of Applied Mathematics and Statistics (AMS) at Johns Hopkins University (JHU). Prior to this, they were a Research Fellow at the Simons Institute at UC Berkeley during the fall 2024 MPG program. Yuqing earned their PhD in Mathematics from Georgia Institute of Technology under the supervision of Prof. Molei Tao.
Yuqing’s research focuses on the mathematical foundations of deep learning theory, particularly from a dynamical systems perspective. Their work bridges machine learning and applied mathematics, employing tools from optimization, stochastic dynamics, computational math, analysis, topology, and sampling. Current interests include large language models and diffusion models.
Yuqing’s publications highlight themes such as training dynamics, large learning rate effects, implicit biases in neural networks, diffusion model design, and optimization techniques. Key contributions include analyses of gradient flow in neural networks, balancing effects in training, and control-theoretic approaches to sampling and attention mechanisms.
As an instructor at JHU, Yuqing teaches Bayesian Statistics and Optimization courses. They have also served as a teaching assistant for differential equations and multivariable calculus at Georgia Tech. Yuqing actively participates in academic talks, with recent presentations at institutions like MPI MiS, UCLA, and conferences such as ICLR, NeurIPS, and SIAM DS25.
