Weak error analysis for stochastic gradient descent via diffusion approximation
Speaker: Yuanyuan Feng, Penn State
Abstract: I will talk about diffusion approximation to stochastic gradient descent algorithms. We prove that diffusion approximation provides weak approximation for stochastic gradient descent algorithms in a finite time horizon. We then introduce new tools motivated by the backward error analysis of numerical stochastic differential equations into the theoretical framework of diffusion approximation, extending the validity of the weak approximation from finite to infinite time horizon.
Room Reservation Information
Room Number: 114 McAllister
Time: 2:30pm - 4:30pm