In order to protect Earth from the threat of asteroid impacts, scientists need to know where they are—and where they're headed.

But while asteroids pose a threat, they also offer opportunities to be mined for precious resources that can help power space exploration. Planetary Resources, a private corporation, is building toward a future in which humans can extract from asteroids both water and rare metals such as platinum, which asteroids may contain in massive quantities.

Together with NASA and the Crowd Innovation Lab at Harvard University, Planetary Resources came to Topcoder to develop an algorithm that discovers new asteroids by analyzing telescope images, yet doesn't increase the number of false positives that create more manual work for scientists.

Topcoder launched a series of 10 crowdsourcing challenges in which data scientists competed to create algorithms that mimic how humans sort through data to discover asteroids. Contestants vied for $35,000 in prizes throughout the project, and a public leaderboard helped drive competition.

At the end of the project, Topcoder delivered a machine learning algorithm 15% more accurate than manual methods of identifying asteroids in the main belt between Mars and Jupiter. Topcoder also developed a desktop application that enables the algorithm to run on everyday laptops—expanding the task of asteroid hunting to the citizens of Earth.

Data Scientists
Boost in Accuracy