Netflix Hopes To Improve Personalisation With Deep-Learning

Netflix Deep Learning

Netflix have used a blog post to detail how the world’s largest on-demand video provider has been experimenting with deep learning. Their methodology has involved using artificial neural networks to provide users with highly specialized content recommendations.

Researchers Alex Chen and Justin Basilico detail how they have been using Nvidia graphic processing units operating in Amazon Web Services’ public cloud infrastructure with the goal of replicating how a viewer’s brain functions.

At this stage it is unclear on what is being planned or how their work will be implemented. The blog’s authors do not go into detail about the specifics of the project beyond saying that it will be used for “personalisation of recommendations”.

Nonetheless, their post is intriguing. Netflix currently streams over two billion hours of video every month and has a global user base of 44 million, meaning it already has a vast amount of knowledge regarding user data and viewing habits. They have information on a person’s preferences for genres, directors, writers and actors that no other TV or film company in the world can match.

If you add this knowledge to their existing physical infrastructure there is no limit to what may be possible for the cloud-based provider. The concept of deep learning inherently requires making inferences based on the data available, whether that is a photo, text, a video or other form of information. If Netflix can effectively harness this data and integrate it into their deep-learning technologies it could revolutionise the world of on-demand video.

Equally notable is the blog’s proof that Netflix is confident in its cloud strategy, despite the loss of prominent cloud-computing expert Adrian Cockcroft last month. Chen and Basilico state, “Implementing bleeding edge solutions such as using GPUs to train large-scale Neural Networks can be a daunting endeavour. 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”.

While the news is undoubtedly good publicity for both Amazon and Nvidia – if the technology can be developed in the way Netflix hope, it will provide them with huge commercial opportunities in the future. By joining Pinterest, Yahoo, and Facebook in the early pursuit of deep-learning technology, the company clearly believe there are huge benefits available and that commercialisation of the field is closer than many would have believed.

By Daniel Price

Gary Bernstein
AI-powered identity verification Even if you don’t want to admit it, doing business online in today’s environment poses a greater risk. Criminals are constantly on the lookout for vulnerabilities to exploit, including hacking, data breaches, ...
Alex Dean
Enabling Privacy and Personalization Most businesses today rely on data collected online to better understand their customers and deliver more personalized products, services and experiences. These insights can be transformative for an organization, especially when ...
Bailey Smith
Intelligent Deals: The Role of AI in M&A It’s no secret that artificial intelligence (AI) is revolutionizing many industries with its fast capabilities and predictive nature. From writing code to drafting documents, AI has become ...
Ronald van Loon
The increasing adoption of technology and AI in business continues to drive concerns regarding sensitive data and the protection of assets. Organizations must implement tools to protect data while also leveraging that data to identify ...
Vulnerabilities
Cyber Threat Intelligence In an era of rapid digital transformation, we have witnessed a concerning evolution in the cyber threat landscape. Recent data analyses, as illustrated in the "Cyber Threat Intelligence Index: Q3 2023" report, ...
Gilad David Maayan
What Is Object Storage? Object storage, in the simplest terms, is a data storage architecture that manages data as objects, as opposed to traditional block storage or file storage architectures. These objects include the data, ...