PhD student, Microsoft Research Asia, University of Science and Technology of China
1 paper at NeurIPS 2025
In this paper, we propose the continuous SGD/SVRG flow of minimizing KL divergence on Wasserstein space