Computational and Information Sciences Directorate Research Areas

Statistical Data Mining

Advisor:  B.A. Bodt
Aberdeen Proving Ground, Maryland

Key words: combat simulation, high dimensionality, supervised learning, data stream

Research is needed to meet persistent challenges in data mining. In one application area, ARL has traction in relating high performance computing, combat simulation, and data mining to support identification of Commander’s Critical Information Requirements. We are concerned with research in statistical knowledge discovery, leading to automatable algorithmic applications to assist the Soldier in utilizing nontraditional high-resolution battle execution measures. Challenges include mining complex data sets requiring multiple simulation executions and characterized by high-dimensional sparse matrices and ordinal/frequency measures. In another area of application, given continued interest in network centric warfare, research is needed to develop tools for managing and inferring from a continuous flow of information over the network. For envisioned decision-making applications, data must be processed continuously and patterns inferred without multiple passes over a static image.