Rare Event Detection in a Spatiotemporal Environment

Abstract

In this paper we explore the use of Extensible Markov Models (EMM) to detect rare events in a spatiotemporal environment. This initial work shows that an EMM is scalable, dynamic, and can detect rare events based on spatial values, temporal values, or transitions between real world states. The core of the EMM approach is a combination of clustering and dynamic Markov Chain. Keywords—Extensible Markov Model (EMM), rare event