Asymmetric Warfare —
STALK (Shaping Tactics by Assessing Knowledge and Learning)
In Iraq, Afghanistan, and other areas of the world, terrorists and insurgents enjoy a significant advantage. They launch surprise attacks against weakly defended targets with improvised explosive devices (IEDs) and other weapons, disappearing before U.S. forces can respond. Because these groups change tactics frequently, the location and means of their next attack are difficult to predict.
To respond to this kind of asymmetric adversary, the military needs a tool that will not only predict where future attacks will occur but also provide insight into enemy thinking. This last criterion is critical if the military is to shift from a reactive to a proactive stance.
Tools for a Static Enemy
Existing predictive methods either assume that areas that have been attacked once will be attacked in the future or simply use the interval between attacks at a specific location to predict the next one. These methods work only for short periods of time because they make no consideration for an adapting adversary. They do not model the subject’s decision-making processes, simply assuming that their preferences are static.
Tools for a Dynamic Enemy
CCRi is building a suite of predictive and analytical tools, called Shaping Tactics by Assessing Learning and Knowledge (STALK) to model the enemy’s decision-making process and predict the location of future attacks. This technique utilizes the Discrete Choice Model (DCM) to generate a prediction based on historical attacks and a wide range of spatial and temporal data that serve as proxies for direct observation and local knowledge.
We perform a sophisticated change detection analysis that considers, among other issues, how spatial and temporal features affect enemy decision-making. Changes in the significance of features — their signature — represent changes in behavior (not just a new place to stage an attack.)
In the final stage of prediction, we model enemy learning by considering enemy performance metrics, such as casualties. Based on these performance metrics and our DCM analysis, CCRi can optimize the enemy’s behavior, producing the best course of action and the one the enemy is likely to adopt.
CCRi Capabilities
DATA FUSION
PREDICTIVE MODELING
CLASSIFICATION