Raytheon tapped for self-evaluating machine learning system

WALTHAM, Mass. (UPI) — Raytheon Co. announced on Monday it has begun work on a machine-learning technology allowing machines to teach machines through artificial intelligence use. The $6 million contract is one of four, valued at a total of $20.9 million, between the U.S. Defense Research Projects Agency and Raytheon BBN Technologies, SRI International, BBN Technologies, Teledyne Scientific & Imaging and BAE Systems. The new deal calls for the development of systems able to communicate information and the conditions of the initial learning, and recommended strategies and situations calling for those strategies. Known as CAML, or Categorical Abstract Machine Language, it uses a process similar to that in a video game; instead of rules, the system offers a list of choices and identification of a goal. By repeatedly playing the game, the system will learn the best way to achieve the goal. CAML focuses on competency awareness machine learning, so that a machine, using artificial intelligence, can self-assess its capability and strategy, and express them in a way understandable to humans. “The CAML system turns tools into partners,” said Ilana Heintz, a principal investigator for CAML at Raytheon. “It will understand the conditions where it makes decisions and communicates the reasons for those decisions.” The four-year DARPA program will involve three years of research and one year of technology demonstration. Reporting by Ed Adamczyk