A new artificial intelligence technology allows U.S. soldiers to learn 13 times faster than conventional methods. The Army researchers who developed this technology say that it could help save many lives on the battlefield.
The development comes from the U.S. Army Research Laboratory. Scientists at this lab focused on improving the rate of learning. They found that it is possible to help soldiers decipher hints of information and deploy solutions faster with a little help from AI.
University of Southern California student Shijie Zhou (center) holds a computer board. ARL West's Dr. Raj Kannan (left) and USC Professor Viktor Prasanna (right). Source: U.S. Army Research Lab
The team used cheap, lightweight hardware with implemented collaborative filtering and field programmable gate array platform for their research. With this system, the team achieved a 13.3-times speedup of training, compared to other state-of-the-art systems. The new system also had 12.7-times speedup for optimized GPU systems. Along with faster learning, the new system uses less power too. The new system used 13.8 watts while other systems use 130 watts for multi-core and 235 watts for GPU platforms. These new components would be incredibly useful for adaptive and lightweight computing systems.
The new technique could be a major part of the next generation of combat vehicles. The technique offers cognitive services and devices for warfighters in war environments. The new technology is one of six Army Modernization Priorities that the scientists in the Army lab are currently focusing on.
The research team working on this project under the Army’s larger focus on artificial intelligence and machine learning. Kannan said his team is developing a few techniques that speed up AI/ML algorithms.
The paper on this new research won the best-paper award at the 26th ACM/SIGDA International Symposium on Field Programmable Gate Arrays in Monterey, California in February of this year.
