Concurrent Agent-enabled Feature Extraction (CAFÉ)

High fidelity, large scale simulations of complex systems pose a very difficult situation to the scientist trying to understand the physical domain and characteristics of their system. Often, it is impossible to manually search through the data that can come out of these simulations, which may range from Gigabytes to Terabytes. Furthermore, it may be impossible to visualize the multi-dimensional interactions that are occurring in the data. A tool is needed that will concurrently data-mine the vast data sets that are produced and alert the scientist of significant events (either planned or unplanned). 21st Century Systems, Incorporated and Brigham Young University introduce our Concurrent Agent-enabled Feature Extraction (CAFÉ) concept to answer this challenge. CAFÉ features state-of-the-art data-mining and analysis leveraging BYU’s expertise. An innovative intelligent agent structure from 21CSI will allow concurrent data-mining utilizing information sharing that will make it possible for multiple analysis methods to work together to improve the data-mining performance. The agent design, specifically the evidential inference engine, also allows direct collaboration with the data-mining algorithms by the scientist. In this way, CAFÉ allows the scientist to observe and correct the data-mining of high-fidelity fluid dynamic simulations maximizing valuable research time. BENEFIT: Searching for physics-related events in complex large scale simulations will be enhanced by a tool that extracts and displays accurate, near real-time patterns from massively large data sets. With CAFÉ, the potential exists to go even further through the interaction of concurrent, on-the-fly, queries and responses among the CAFÉ agents as well as with the human operator within a single social network. During Phase I, with the approval of the AFOSR TPOC, 21CSI will contact the Air Force Research Laboratory (AFRL) Information Directorate at Rome Labs to acquire end user requirements for CAFÉ. The 21CSI team fully intends to build and transition key technological elements such as the collaborative agent concept and information inferencing products using the collaborative environment derived from 21CSI’s Evidential Reasoning NetworkTM (ERN) technology – to suitable USAF and DoD development centers for concurrent data mining. Furthermore, military command & control (C2) operations involved with blast and fragmentation effects in urban areas will benefit from this technology. Urban combat damage to structures could inhibit troop movements or induce sniper stands. The Battle Command Battle Lab - Ft. Huachuca is interested in the technology for a new training simulator for artillery planners and commanders. CAFÉ will expand the capabilities already developed for Ft. Huachuca by allowing us better insight into blast events and produce a more accurate simulation. The team of 21CSI and BYU will pursue further development of CAFÉ with proposals to General Electric Aircraft Engines. Currently, it costs more than $2 billion and 10 years to develop a new aircraft engine. BYU’s vision is to create a world-class virtual simulation of an entire jet engine to allow repetition in the process to be done virtually at a lower cost and shorter time. With the addition of CAFÉ the experimentalist can perform concurrent efforts saving simulation time and data storage costs. Together, this would help to uncover problems before a significant investment is made in building a physical jet engine model. While physical testing can never be eliminated, it can be reduced from current set of 12 versions of development engines – a costly and time-consuming process. Another U.S. government, but non-DoD, area of active interest is modeling to stop terrorist activity here at home. At the Lawrence Livermore National Laboratory (LLNL), the Nonproliferation, Homeland and International Security (NHI) directorate aims to be the premier provider of systems solutions to preclude WMD proliferation from terrorist attack. LLNL will partner with research institutions and industry to transition new capabilities to operational use. Mitigating the consequences of potential effluent plumes from a released WMD that a terrorist might want to use at a large public gathering coupled with [un]favorable weather might be desirable to the evil-doers. Countermeasures include modeling the combination of potential plumes and weather forecasts to provide best use of civilian law enforcement and US government personnel.