报告题目： A framework for probability approximations

报告人：徐礼虎 （澳门大学）

报告时间：4月7日 10:00-11:00

报告地点：管理楼1418

摘要：

By embedding the classical Lindeberg principle into a Markov process and using conditional expectation, we establish a general probability approximation framework. As applications, we study the error bounds of the following three approximations: approximating online stochastic gradient descents (SGDs) by stochastic differential equations (SDEs), approximating stochastic variance reduced gradients (SVRGs) by stochastic differential delay equations (SDDEs), and the approximation of ergodic measure of stable SDEs by Euler-Maruyama scheme. More applications will be discussed. This talk is based on the joint works with P. Chen, J. Lu and Q. M. Shao.