FIRESTORM: Framework for Information Reasoning and Extraction for STatistical Object Recognition and pattern Matching

The trend toward storage of data collected by the U.S. military and Intelligence Community (IC) in high performance computing facilities (i.e.,"data clouds") enables historical retrospective comparison of data to new collections to facilitate pattern discovery for prediction. However, many of the techniques for reasoning and pattern recognition have yet to be applied to the emerging data cloud technologies. Moreover, the diversity of information now supported by data stores and computing scales of"Internet"magnitude exposes new opportunities for analysis, situational awareness, prediction and planning. Harmonia"s Framework for Information Reasoning and Extraction for STatistical Object Recognition and pattern Matching (FIRESTORM) is designed to assist the warfighter and intelligence analyst to evaluate massive datasets across many different dimensions, to find patterns that may predict critical impacts to operations, or windows of opportunity for mission courses of action. FIRESTORM will use DCGS-N Increment 2 and DCGS-A IC ITE technologies, including open source"Storm"topologies, to continuously transform data ingested by a cloud into layers that can be mapped to multiple dimensions of time and geography; and to automatically recognize patterns, using user-driven case-based reasoning, via massively parallel computation with GPU algorithms, Hadoop MapReduce, and Harmonia"s ONR-funded BumbleBee to coordinate computations across clouds.