Title: Improving Collection with Forecasting
Program Manager, NGA
National Geospatial-Intelligence Agency
I will discuss recent work using mathematical models to understand and predict event data ??? data labeled with a place, time, and type of activity. The work builds on Self-Exciting Point Process (SEPP) models used for pattern analysis and prediction in seismology and crime prevention. In addition to giving intuitive descriptions of different types of point processes, I will show how to use them to forecast, and how to evaluate those forecasts. One of the applications for this work is to drive image collection by satellites. I will also give a brief overview of the National Geospatial-Intelligence Agency (NGA) and some of its biggest scientific challenges.