Advertisment

MIT researchers can predict future with an algorithm

author-image
CIOL Writers
New Update
CIOL MIT researchers can predict future with an algorithm

In what may sound like a chapter straight out of a science fiction novel, researchers at MIT's Computer Science and Artificial Intelligence Laboratory are actually teaching computers to see into the future. They've created an algorithm that can forecast hugs, kisses, and high-fives, even before they happen in reality.

Advertisment

The algorithm can look at a photo of two people and predict what's going to happen next. For instance, if I showed this software a photo of you and me meeting on the street, it can anticipate whether we're likely to hug, kiss, and shake hands or high-five.

CIOL MIT researchers can predict future with an algorithm

The algorithm uses artificial intelligence, deep learning, and computer vision—few of the most groundbreaking technology innovations of recent times. To feed the computers with a lot of examples, and to train the computers to recognize the human behavior, researchers turned to a massive collection of recorded human behavior on YouTube. MIT researcher Carl Vondrick said they had downloaded 600 hours of video from YouTube for the purpose--some from TV shows like Desperate Housewives and The Office, others were user-generated. With each clip, the intelligent computer learns something new about human behavior.

Advertisment

"So it's learning, for example, that when someone's hand is outstretched, that means a handshake is going to come," Vondrick said. "Or maybe that two heads are very close together, that might suggest a kiss."

Once the researchers trained the algorithm, it could make predictions or forecasts a few seconds in the future. So they could pause a video, and ask the algorithm — what do you think is going to happen next? Will they shake hands? Will they kiss? Will they hug? Or will they high-five?

The system is currently about 43 per cent accurate, but not nearly as good as human beings. Vondrick, however, says that the system will get much more accurate as they feed more complex data in it to study human behavior.

In its current form, the system would not me much of help in critical situations of security, but with enough training data, the algorithm could be used to predict any kind of human behavior. Vondrick said he's particularly excited about potential health care applications.