The Robot Reckons

researchers, from left, Ruslan Salakhutdinov, Brenden Lake, Joshua Tenenbaum Alain Decarie for The New York Times

Another barrier separating human from automaton abilities has been breached. Researchers at MIT, NYU, and the University of Toronto have developed a program that outperforms humans in the ability to quickly identify handwritten characters based on a minimal number of examples. The program uses B.P.L., or Bayesian Program Learning, which is different from the deep neural network learning technology currently used. It is able to learn the characters after seeing only one or two examples and to generalize from there, which, the researchers say, is similar to the way in which humans learn. "We are still far from building machines as smart as a human child," notes MIT's Joshua Tenenbaum, "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": http://www.ctvnews.ca/sci-tech/new-algorithm-lets-machines-learn-like-humans-1.2695230