Mathematics - Data Science Seminar, Arnab Ganguly, A framework for nonparametric inference of schoc
This event is in the past.
Speaker: Arnab Ganguly, Louisiana State University
Time: Wednesday, December 6, 1:30pm-2:30pm
Place: Virtual
Zoom link:
https://wayne-edu.zoom.us/j/96316494795?pwd=Ylc3M0R0R1BYaUZGSnB2dkI2UFRVQT09
Title: A framework for nonparametric inference of stochastic differential equations
Abstract:
Stochastic differential equations (SDEs) provide powerful tools for modeling temporal evolution of a variety of systems. Statistical estimation of these systems address finite-dimensional inference problems, presuming that the stochastic models are fully known up to a finite number of real-valued parameters. Although the assumption of known functional forms of the driving functions is effective in certain scenarios, a considerable number of realistic systems require a more data-driven approach, where these functions are entirely learned from available data. These problems fall in the realm of infinite-dimensional learning in function spaces. In this talk, we will explore an approach that integrates RKHS theory and Bayesian methods to develop learning algorithms for these problems. If time permits we will discuss the cases of both high-frequency and sparse and noisy datasets.