题目：New algorithms for samplings and diffusion models
报告人：张希承
报告时间：2024年6月20日15:30-16:30
地点：综合楼644会议室
报告人简介：
北京理工大学数学与统计学院教授。研究方向主要为随机分析及其应用。曾获国家自然科学基金杰出青年项目资助以及国家高层次人才支持计划。目前在国际期刊发表一百余篇学术论文。
报告摘要：
Drawing from the theory of stochastic differential equations, we introduce a new sampling method for known distributions,
as well as a new algorithm for diffusion generative models with unknown distributions. Our approach is inspired by the concept of the reverse diffusion process, which has been widespread adoption in diffusion generative models. Furthermore, we compute the explicit convergence rate based on the smooth ODE flow. Numerical experiments exhibit the effectiveness of our method. In particular, unlike the traditional Langevin method, our sampling method does not require any regularity assumptions about the density function of the target distribution. We also apply it to the optimization problem.
