search menu icon-carat-right cmu-wordmark

A Series of Unlikely Events: Learning from Sequential Behavior for Activity-Based Intelligence and Modeling Human Expertise

October 2019 Presentation
Eric Heim

This presentation describes work to use inverse refinforcement learning techniques to perform activity-bassed intelligence.

Publisher:

Software Engineering Institute

Abstract

The Department of Defense (DoD) and the intelligence community (IC) frequently analyze activity based intelligence (ABI) to inform missions about routine patterns of life (POL) and unlikely events that signal important changes. We propose an alternative approach, inverse reinforcement learning (IRL), that observes all states and actions in data and computes a statistical model of the world that includes whether each behavior is part of a routine.