Raytheon at work on DARPA-funded new ML program


A prominent U.S.-based defense contractor is working on a system to “create trust” between humans and their robot brethren, with a $6 million contract from the Defense Advanced Research Projects Agency (DARPA) to dig into machine learning technology.

Reports today show that the project will aim to develop new communication models for systems that can learn on the fly.

Descriptions of the project in the press also reveal that Raytheon’s Categorical Abstract Machine Language or CAML system will utilize the same kind of game theory inherent in new initiatives like Q-learning, where the system moves through scenarios and learns to achieve the best outcomes based on Markov decision models.

Here’s a description as posted today at Army-technology.com:

“The system makes use of a process that is similar to that in a video game, offering a list of choices and identifying a goal instead of rules. It will repeatedly play the game and learn the most effective way to achieve the goal. It will also record and explain the conditions and strategies used to come up with successful outcomes.”

Raytheon’s bid isn’t the only U.S. defense project to utilize advanced technology – for instance, the U.S. Army already has a virtual recruiter and information liaison called “Sergeant Star” that is online already. (whether Sgt. Star passes the Turing test is for you to find out – go check it out.)

But the new machine learning project represents a rare public/private investment into what private businesses have been pursuing for several years – self-learning programs based on probabilistic programming, instead of linear deterministic programming principles.

It will be interesting to see what the bureaucratic combination of grant funders and U.S. companies and defense agencies come up with, but one thing’s for sure – machine learning is moving ahead quickly.

“The rate of progress in the field of artificial intelligence is one of the most hotly contested aspects of the ongoing boom in teaching computers and robots how to see the world, make sense of it, and eventually perform complex tasks both in the physical realm and the virtual one,” wrote Nick Statt last December at The Verge. “And just how fast the industry is moving, and to what end, is typically measured not just by actual product advancements and research milestones, but also by the prognostications and voiced concerns of AI leaders, futurists, academics, economists, and policymakers.”

Keep an eye on this innovative ML model to see whether new advances come out of the defense industry or somewhere else.