Newton-Type Stochastic Optimization Algorithms For Machine Learning
计算数学学术报告
报告题目(Title)：Newton-Type Stochastic Optimization Algorithms For Machine Learning
报告人(Speaker)：文再文 （北京大学）
地点(Place)：教八410
时间(Time)：2019年9月23日13:30
邀请人(Inviter)：陈华杰
报告摘要
In this talk, we introduce 1) stochastic semismooth quasi-Newton methods for large-scale problems involving smooth nonconvex and nonsmooth convex terms in the objective function for deep learning; 2) a stochastic trust region method for reinforcement learning.
