Small Data is (and may always be?) the Problem

Francis J. Alexander -- Brookhaven National Laboratory
Cels Seminar Graphic

In this talk Francis J. Alexander will discuss the challenges of having insufficient data for problems involving complex systems. Alexander will also describe some of the methods his colleagues and him have been developing to overcome these challenges.  These methods include transfer learning, the development of optimal priors, and optimal data acquisition.  Future directions including wicked problems and resource constrained optimization will also be discussed.