EcoSta 2021: Start Registration
View Submission - EcoSta2021
A0564
Title: Drift estimation for a multi-dimensional diffusion process using deep neural networks Authors:  Yuta Koike - University of Tokyo (Japan) [presenting]
Akihiro Oga - University of Tokyo (Japan)
Abstract: Recently, many studies have shed light on the high adaptivity of deep neural network-based estimators in the framework of nonparametric regression. In particular, their superior performance has been established for various multivariate function classes. Motivated by this development, we propose estimating the drift coefficient of a multi-dimensional diffusion process by deep neural networks from its discrete observation data. Then, we derive their generalization error bounds and show that they achieve the minimax estimation rate up to a logarithmic factor.