Data-driven Intelligence and Security
The Data-driven Intelligence and National Security group applies innovative approaches to issues of national security. Using a cross-disciplinary approach combining computer science, data science and novel analysis techniques, our group can discover patterns and phenomena often overlooked when using traditional approaches. As an example, fusing heterogeneous data sets often leads to new insights, patterns and discovery of new phenomena. These deep learning techniques are often discussed, however, execution with regard to specific problems is not trivial. This is especially true when data sets are streaming, massive and created to be disparate.
Our group focuses on foundational technology that underpins deep learning. This includes hardware integration (e.g., FPGA, GPU), and algorithm development (e.g., artificial intelligence, neural networks, Markov models, and Bayesian Belief Networks). Novel approaches are applied to both streaming and resting data. Using strict modulization for code, our foundational libraries can be addressed to myriad specific and practical applications.
The group maintains a “sandbox node” with hardware accelerators and associated libraries as well as libraries of analytical code, which is available to the Brookhaven Lab research community.