There’s a trend with the online giants like Google and Facebook, to create algorithms that resemble the way a human brain works in order to better serve up search queries or photo tagging. Netflix is also getting involved, but in a different way. The streaming movie service has always done a better than average job of recommending films and shows to users based on their habits. It started with Netflix relying on what your friends had watched to suggest things to you, because friends make the best movie recommendations.
Since then Netflix has gotten more automated, but now wants to approach the idea of the personal recommendation from a new angle. They’re going to get into “deep learning,” a system that mimics the neural pathways in the human brain to create artificial intelligence. Rather than build their own array of physical computers and servers, Netflix is turning to the Amazon cloud system. They’ve written the new algorithms that utilize graphical processing units in order to complete the computations, but will upload them to the Amazon cloud.
How they plan to implement the new system is unclear, or even when it’ll be released. The idea is still in the early stages. It is an open source system, and hopefully by making this leap onto the Amazon cloud other companies and brilliant people will follow suit. Thus making the “deep learning” corner of artificial intelligence viable for those who don’t have billions of dollars at their disposal.
At the least we can hope that the recommendations made by Netflix when we log in will get better. Just because I like Kurosawa does not mean I want to watch 3 Ninjas. Netflix has detailed their research, work, and coding in a recent blog post that is a dense read. It is fascinating to tap a bit into the minds of those who are making such progress in artificial intelligence, all so we can have better movie recommendations.
Staff Writer at CinemaBlend.
Your Daily Blend of Entertainment News
Thank you for signing up to CinemaBlend. You will receive a verification email shortly.
There was a problem. Please refresh the page and try again.