Netflix moves into deep learning research to improve personalization

Google isn’t the only company looking to make lots and lots of computers function more like human brains.

In a blog post published earlier today, Netflix research engineers Alex Chen, Justin Basilico, and Xavier Amatriain talked about ways the movie-streaming company has been experimenting with the construction of artificial neural networks using Nvidia graphic processing units (GPUs) running in Amazon Web Services’ vast public-cloud infrastructure. In other words, these new methods could do things like improve your video recommendations by mimicking how your brain works.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":893317,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,","session":"B"}']

The blog post doesn’t explain the specific purposes of the research, only that it’s for “personalization” of recommendations for each user. The company does, however, go into detail about specific experiments with different types of GPUs and training efforts.

Even if we don’t know exactly what’s being planned, it’s significant when a company like Netflix — which has lots of users, lots of data, and lots of infrastructure to tap — enters the emerging, hype-heavy realm of deep learning. Such work entails making inferences based on the data that’s available, whether it be a photo, a video, text, or something else. It also suggests commercial value (aka video recommendations) could be closer than we might think.

But the news is notable for a few other reasons, too. It shows just how confident Netflix is in its all-in cloud strategy, which executives at Amazon Web Services love to talk about. If companies want to use GPUs for deep learning, Amazon has it, and now Amazon can talk about a neat new way that companies are using its GPU servers.

“Implementing bleeding edge solutions such as using GPUs to train large-scale Neural Networks can be a daunting endeavour,” the post’s authors wrote. “If you need to do it in your own custom infrastructure, the cost and the complexity might be overwhelming. Levering the public AWS cloud can have obvious benefits, provided care is taken in the customization and use of the instance resources.”

The research also is great for Nvidia, because it shows a big-name company using its GPUs. Other companies that have looked at deep learning or have brought on talent for it recently include Pinterest, Yahoo, and Facebook. Now there’s another big webscale company looking into the field.