Netflix aims to improve recommendations via a distributed neural network

Feb 14, 2014
2
Netflix aims to improve recommendations via a distributed neural network

If you are a Netflix subscriber, you have undoubtedly watched some of the films that the recommendation engine lists for you at one point or another. I'd wager you have also looked at some of those films and wondered exactly how Netflix thinks they are related to what you like. Netflix is always out to make that recommendation service better.

The company knows that the current system often adds too many films that people never watch. One of the ways that Netflix is trying to improve its recommendation service is with the development of a new algorithm that uses deep learning techniques.

Netflix calls its plan a Distributed Neural Network and it will use cloud computing power from Amazon Web Services along with GPUs from NVIDIA to power the improvements. Using the tech, Netflix has trained its cloud to make personalized recommendations in 47 minutes. Previously that recommendations process took 20 hours.

It's a good thing for users that Netflix is working on tweaking how it recommends movies. Better recommendations will mean more stuff for you to watch and fewer questionable recommendations that you might not like.

SOURCE: TechRadar


Must Read Bits & Bytes