When you want to watch a movie, nothing ever matches the advice of a friend who knows your tastes, but that could change in the near future! Netflix is developing a unique kind of online search engine that understands and predicts your cinematic tastes better than your loved ones.
Netflix, the online video broadcasting company has already set up the Netflix Prize, an algorithm to better predict movies and TV series that matches your taste. The company has now gone a step further, in order to improve the search engine recommendations, it has hired scientists in the field of “deep learning”, a branch of artificial intelligence specializing in solving complex problems using computers to mimic the structure and behavior of the human brain.
Through this project, Netflix is following the footsteps of other Internet giants such as Google and Facebook, which have also hired leading researchers in “deep learning” (deep learning) to improve speech recognition and image tagging. However, unlike google and Facebook the Netflix goes about it a little differently. Rather than create its own infrastructure, the company plans to use the Cloud of Amazon to develop its algorithms. In this way, the small companies can compete the bigger giants.
Many methods of “deep learning” use neural networks to simulate the working of human brain. Owing to the fact that we do not completely know the working of human brain and the utility of these methods is limited, still it can be a good start. It has been only very recently possible to use the “deep learning” techniques for practical problems as computers have become powerful enough now. Even then, according to Engineering and Technology Magazine the best neural networks can simulate only 1% of the human brain. Recent advances in “deep learning” can detect faces in photos, learn your tastes and habits, and to some extent, understand what you are saying.
A wide range of companies are using such systems: Microsoft, Google, Facebook, also Baidu, the Chinese search engine and now Netflix. A key researcher at the heart of this research is an academician named Geoffrey Hinton, working at Google.
Most of today’s neural networks operate on the graphics processors, that are designed to handle the processing of high-quality images very quickly. These processors were originally designed for computer gamers and mechanical designers. Currently, these processing units are grouped together as “farms” where a large number of these processors work in parallel to run all kinds of calculation-intensive applications, including the “deep learning” algorithms. Engineers at Netflix ensure that the “deep learning” is now economically viable, mainly due to graphics processors invented by Andrew Ng, a professor at Stanford and a researcher at Google.
Netflix, just like Google uses these graphics processors, but the company will not build its own farms, since it will rely on the Amazon’s Cloud . The company did not disclose further details about how neural networks are applied to make personal recommendations work. The work is still at an experimental stage although apparently feasible in the very near future. This could certainly give ideas to other small businesses and even individual researchers could also use Amazon Cloud to conduct their experiments.
This research will prove to be very exciting, although it is still in its infancy at the moment. In a few years, we may interact with our computers so that they can serve us better. What is your opinion about this new search engine of its kind?