A somewhat unexpected application of deep neural networks is to use them to solve partial differential equations (PDE). Especially for very high-dimensional PDE, where classical methodology suffers from "the curse of dimensionality", the deep learning approach has proven to be very competitive. This is explored in this project in order to solve stochastic PDE related to the nonlinear filtering problem.
Links related to the project
Deep learning for solving PDE and for stochastic control
(lecture by Adam Andersson)
Recruitment