Extension of probability flow on matrix Lie group
组织者
演讲者
时间
2023年01月17日 21:30 至 22:00
地点
Online
摘要
Feedback control for the probability flow is firstly proposed by Prashant G. Mehta et al. Given a smooth path ${p_t^*in mathcal{P}_2(mathbb{R}^d)}$, basic problem is to design stochastic process ${{X}_t}$ such that the probability density ${p}_t$ equals to $p_t^*$ for any time $tge 0$. Note that it acts as a general framework. Solution of probability flow is of interests in many applied areas. Up to 2021, there exist complete formulation under Euclidean space. In this talk, we will discuss the extension to matrix Lie group. We will discuss two cases: deterministic and stochastic probability flow.
演讲者介绍
焦小沛,本科毕业于上海交通大学致远学院,博士毕业于清华大学数学科学系。先后在北京雁栖湖应用数学研究院,荷兰特文特大学从事博士后工作。现研究方向包括有限维滤波理论,丘-丘滤波方法,物理信息神经网络以及生物信息学。研究兴趣主要集中于(1)利用李代数等几何工具进行偏微分方程求解与有限维滤波系统的分类;(2)设计基于物理信息神经网络的新型数值算法。