DARPA chooses teams to develop approaches to machine learning, artificial intelligenceNews
May 09, 2018
ARLINGTON, Va. The Defense Advanced Research Projects Agency (DARPA) has selected teams to work on its Lifelong Learning Machines (L2M) program, which the agency says seeks to develop fundamentally new machine learning (ML) approaches that allow systems to adapt continually to new circumstances without forgetting what they have previously learned.
The L2M research teams are now applying their diverse expertise toward understanding how a computational system can adapt to new circumstances in real time and without losing its previous knowledge. One group -- located at the University of California, Irvine -- plans to study the dual memory architecture of the hippocampus and cortex and apply that knowledge to create an ML system capable of predicting potential outcomes by comparing inputs to existing memories. Such a system would theoretically become more adaptable while retaining previous learnings.
Another group, headquartered at Tufts University, is examining a regeneration mechanism observed in animals (in salamanders, for example) to create flexible robots that are capable of altering their structure and function on the fly to adapt to changes in their environment. The team working at the University of Wyoming, for its part, will adapt methods from biological memory reconsolidation to develop a computational system that uses context to identify appropriate modular memories that can be reassembled with new sensory input to rapidly form behaviors to suit novel circumstances.
“With the L2M program, we are not looking for incremental improvements in state-of-the-art AI [artificial intelligence] and neural networks, but rather paradigm-changing approaches to machine learning that will enable systems to continuously improve based on experience,” said Dr. Hava Siegelmann, the program manager leading L2M. “Teams selected to take on this novel research are comprised of a cross-section of some of the world’s top researchers in a variety of scientific disciplines, and their approaches are equally diverse.”
Siegelmann adds: ““We are on the threshold of a major jump in AI technology. The L2M program will require significantly more ingenuity and effort than incremental changes to current systems. L2M seeks to enable AI systems to learn from experience and become smarter, safer, and more reliable than existing AI.”