If you can teach computers to learn, NASA needs your help.

Sometimes the distortions are obvious, like in the Hubble image of a distant galaxy cluster above. But sometimes they're too subtle to be picked out by human eyes, and can even be confused with noise from the telescope used to take the galaxies' picture. So cosmologists have turned to machine learning algorithms that teach computers to recognize patterns.

"We're trying to teach computers to pick out the correct shape given all sorts of other noise around the galaxy's shape," said NASA cosmologist Jason Rhodes, who is helping to organize the challenge. "We have our ideas as a community about how to do this, but we realized a few years ago that it was quite possible there were ideas we weren't familiar with."

The competition is designed to bring fresh ideas from machine learning and computer science experts. But the challenge is open to anyone.

Full details at Wired