Cognitive System Accelerator of Knowledge Gain in Human-computation-sensor Analytic Systems
Small Business Information
5050 Section Avenue, Suite 110, Cincinnati, OH, -
AbstractABSTRACT: In an attempt to address the data overload challenge significant work has been conducted studying the human factors and cognitive science issues related to exploitation of"big data". Also, significant investment has been made in automated or autonomous processing/exploitation capability. It is clear that no matter how efficiently the human cognitive ability is applied to"manually"interpreting data that we cannot keep pace with the data explosion in this manner. In spite of the allure of automated or autonomous processing/exploitation the"brittleness"of current capabilities limits the benefit achieved because significant human oversight and correction is required What is required to meet this throughput/brittleness challenge is focus on the human-machine system and, in particular, the development of adaptive human-machine systems that evolve, learning from each other to continually improve system performance and accurately adapting the roles (boundaries of competency) of each agent in order to optimize performance of the total human-computation-sensor system. The results of our Phase I research highlighted opportunities for cognitive acceleration of knowledge gain across the various points of collection, processing, interpretation and transmission of both human and automated analysis activities. This work points to potential order-of-magnitude improvements in throughput and quality. The proposed Phase II research will implement and assess these concepts by addressing data exploitation challenges associated with a major new hyperspectral program. BENEFIT: The proposed work has the potential to achieve order-of-magnitude increases in throughput and quality of integrated intelligence production systems. This will help the ISR community to deal with a rapidly growing deluge of data from an increasingly diverse and disparate set of sensors and sources, within increasingly stringent time requirements. This work will benefit NASIC, DoD, NGA and other intelligence production organizations and will ultimately address civilian"big-data"applications.
* information listed above is at the time of submission.