Machine learning for personalization

For many years, the main goal of the Netflix recommendation system has been to get the right titles in front of each member at the right time. Tony Jebara details the approaches Netflix uses to recommend titles to users and discusses how the company is working on integrating causality and fairness into many of its machine learning and personalization systems.

For many years, the main goal of the Netflix recommendation system has been to get the right titles in front of each member at the right time. For instance, the 2006 Netflix Challenge helped spur new research in low-rank matrix decomposition and collaborative filtering. Today, the company uses nonlinear, probabilistic, and deep learning approaches to make even better rankings of movies and TV shows for each user. But the job of recommendation does not end there. The home page should be able to convey to the member enough evidence of why a recommended title is a good choice for her, especially for shows that the member has never heard of. One way to address this challenge is to personalize the way Netflix portrays the titles on its service. Its image personalization engine is driven by online learning and contextual bandits to reliably handle over 20 million personalized image requests per second. Finally, while machine learning is great at learning to make accurate predictions, predictions must be made in order to take actions in the real world. Tony Jebara explains how the company is working on integrating causality and fairness into many of its machine learning and personalization systems.