Xiang Cheng

Xiang Cheng is an Assistant Professor in the Department of Electrical and Computer Engineering at Duke University. He holds a Ph.D. from the University of California, Berkeley, and his research interests include machine learning and advanced topics in electrical and computer engineering. Dr. Cheng teaches courses such as ECE 899: Special Readings in Electrical Engineering, ECE 790: Graduate Special Topics in Electrical and Computer Engineering, ECE 590: Advanced Topics in Electrical and Computer Engineering, ECE 580: Introduction to Machine Learning, and ECE 391: Projects in Electrical and Computer Engineering. His profile page can be accessed at https://ece.duke.edu/people/xiang-cheng/.

Xiang Cheng is a researcher at Duke University with expertise in stochastic processes, machine learning, and the theory of deep learning. His work focuses on the theoretical foundations of Markov Chain Monte Carlo (MCMC) methods, particularly Langevin dynamics, and their applications in optimization and sampling. Cheng's research explores convergence rates, non-asymptotic analysis, and the interplay between sampling and optimization in deep learning models. He has also investigated the role of attention mechanisms in in-context learning and the use of Riemannian geometry for efficient sampling on manifolds. His contributions span both theoretical and applied aspects of probabilistic modeling and machine learning.