Computers Learn 'Learning', Take A Step Towards Human Intelligence
Artificial intelligence just got smarter. Researchers report development of an algorithm that help computers learn like humans, with limited data and in a short time.
The Washington Post reports that the algorithm helped computers to recognize letters from a limited set of examples. This was achieved through a process known as Bayesian Program Learning (BPL). When tested against humans, computers had a 3.3 percent error rate in recognizing hand-written letters.
"Our results show that by reverse engineering how people think about a problem, we can develop better algorithms. Moreover, this work points to promising methods to narrow the gap for other machine learning tasks," Brenden Lake of New York University the paper's lead author said in a news release.
Mimicking human learning has been endeavor for many decades but success has been elusive. The recent study presented in the journal Science claims to be the first step towards a far goal. The algorithm developed by Lake and other authors was also able to teach a machine to learn and apply concepts from previous learning efforts. Researchers could make computers learn letters of Greek alphabet through concepts obtained when learning the Latin alphabet.
"Before they get to kindergarten, children learn to recognize new concepts from just a single example, and can even imagine new examples they haven't seen," notes Joshua Tenenbaum, a professor at MIT.
"I've wanted to build models of these remarkable abilities since my own doctoral work in the late nineties. We are still far from building machines as smart as a human child, but this is the first time we have had a machine able to learn and use a large class of real-world concepts -- even simple visual concepts such as handwritten characters -- in ways that are hard to tell apart from humans."