School of Mathematics

Michal Branicki awarded an ONR Grant

Michal Branicki has been awarded an ONR grant, by the Applied and Computational Analysis Division, to research "Geometry of information flow and uncertainty quantification for robust neural network architectures in deep learning".

This project will focus on constructing an information-geometric framework for analysing the learning capacity of deep neural network architectures, quantifying uncertainty and robustness of the estimates, and determining probability of large errors in network output subject to uncertain and incomplete information. The duration of this project is three years.