As part of its Deep Learning program, the Defense Advanced Research Projects Agency (or DARPA) is exploring recent breakthroughs in the ability of machines to learn and assess places and objects.
The need for such research is being driven by the vast amount of data that’s become available to Defense commanders and analysts from new types of sensors. For warfighters, the data has to be quickly and correctly analyzed. Currently, that’s done by highly trained human operators. But as sensor capabilities expand, DARPA says sophisticated, powerful machines with the ability to imitate, and even surpass, human perceptual capabilities will be needed.
They’re building applications that will allow computers to detect and classify objects and activities. So far, the results hold promise for achieving human-level-or-better analysis.