NEWS
NEWS
NEWS
Facebook’s AI can tell you what’s in your photos
Facebook has come a long way since the days of being a hookup site for college kids. The social network does a lot of other things now too, including groundbreaking artificial intelligence research that focuses on creating computer programs that can intelligently answer questions about text and images.
Today, the company revealed that Facebook’s AI Research (FAIR) team has made significant progress on some of its projects, especially when it comes to programs that analyze images.
“Next month FAIR will be presenting a new paper at NIPS, a leading artificial intelligence conference,” Facebook Chief Technology Officer Mike Schroepfer said in an announcement. “In the paper the team details a state-of-the-art system that segments, or distinguishes between, objects in a photo. This new system segments images 30 percent faster, using 10x less training data, than previous industry benchmarks.”
Facebook had previously been working on an AI system called Memory Networks (MemNets) that could answer questions about short texts, and FAIR was recently able to combine MemNets with its image recognition technology to create an AI that can intelligently answer questions about images.
In a video demo of the new system, which FAIR calls Visual Q&A (VQA), Facebook demonstrated that the system was able to correctly recognize that a baby was in the photo, that the picture was taken in the bathroom, and that the baby was having his teeth brushed. While this could certainly be useful for photo tagging features used by Facebook and Instagram, it also could also have major implications for accessibility programs for people with poor vision.
Facebook said that it has also made progress with one of AI’s most notorious challenges: playing the Chinese board game Go. Unlike chess, which was famously conquered by IBM with Deep Blue in 1996, Go is an incredibly difficult game for an AI to play, and there are currently no programs capable of challenging the world’s top players.
Through a combination of machine learning and pattern recognition, Facebook says that its Go-playing robot has taught itself to play as well as other AI-powered systems in just a few months, and it is currently equivalent in skill to “a very strong human player.”
Image courtesy of Facebook
Since you’re here …
… We’d like to tell you about our mission and how you can help us fulfill it. SiliconANGLE Media Inc.’s business model is based on the intrinsic value of the content, not advertising. Unlike many online publications, we don’t have a paywall or run banner advertising, because we want to keep our journalism open, without influence or the need to chase traffic.The journalism, reporting and commentary on SiliconANGLE — along with live, unscripted video from our Silicon Valley studio and globe-trotting video teams at theCUBE — take a lot of hard work, time and money. Keeping the quality high requires the support of sponsors who are aligned with our vision of ad-free journalism content.
If you like the reporting, video interviews and other ad-free content here, please take a moment to check out a sample of the video content supported by our sponsors, tweet your support, and keep coming back to SiliconANGLE.
