An algorithm that can compare two images and that outperforms humans in assessing situations has been developed by researchers from the Computer Science and Artificial Intelligence Laboratory at the Massachusetts Institute of Technology.

Humans have the capability to make assumptions based on the environment. Decisions people make need them to look beyond their "visual scene," and figure out the details in a complex set of understandings and judgments made in real-time.

Like Us on Facebook

The team of researchers crafted an online test to test if their participants could find the closest McDonald's with the least number of steps .

Humans are usually better at the task than the team's algorithm. But the scientists found the computer steadily outperformed humans at a version of the mission in which participants are shown a pair of pictures showing McDonald's branches and asked to pick the nearest one.

The team created the algorithm by training the computer on a set of 8 million Google images across eight prominent cities in United States with global positioning system data on McDonald's locations and crime rates.

The team used deep-learning techniques to help the program teach itself how the distinct attributes of the images correlate. The algorithm, for example, discovered on its own that items commonly found near McDonald's locations include prisons, police vans and taxis.

"These sorts of algorithms have been applied to all sorts of content, like inferring the memorability of faces from headshots," said PhD student Aditya Khosla.

"But before this, there hadn't really been research that's taken such a large set of photos and used it to predict qualities of the specific locations the photos represent."

TagsAlgorithm, computer, computer algorithm, Google street view, Directions, Mcdonalds, Crime Rate