Machine Learning for Recommendation and Personalization
A decade ago, Netflix launched a challenge to predict how each user would rate each movie in a catalog. This accelerated the science of machine learning and matrix factorization. Since then, our learning algorithms and models have evolved with multiple layers, multiple stages, and nonlinearities. Today, we use machine learning to rank a large catalog, i.e. personalized content selection that determines how relevant each of our titles is to you. We also find how to best present these movie choices, i.e. personalized image selection that shows the best imagery from each movie that would most appeal to you. We do this through statistically-principled exploration and optimization of both user engagement and retention.