How are researchers able to create videos of the future?

Answer: by using a deep-learning artificial intelligence system

by News Staff / November 29, 2016
Given a still image from a scene, the MIT Computer Science and Artificial Intelligence Laboratory team's deep-learning algorithm can create a brief video that simulates the future of that scene. Carl Vondrick/MIT CSAIL

A golfer's swing, a wave breaking on the beach and a train rolling in — most of us can imagine how these scenes play out. And now, machines are learning to do the same thing. 

Researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL), have developed a deep-learning algorithm that when given a still image from a scene can create a short video of the seconds that follow, according to the institute.

The CSAIL research team says that future versions could be used for improved security tactics and safer self-driving cars, to name just a few applications. And according to CSAIL Ph.D. student Carl Vondrick, the algorithm can also help machines recognize people’s activities without expensive human annotations.

“These videos show us what computers think can happen in a scene,” Vondrick told MIT News. “If you can predict the future, you must have understood something about the present.”

Platforms & Programs