Recommendations at YouTube
Recommendations at YouTube
Series: Tripods
Location: ENR 2 S230
Presenter: Dr. Trevor Potter, Google
We'll give a high level overview of the YouTube recommendation system. Along the way we'll touch on recurrent neural networks to model users' trajectories, reinforcement learning for modeling long-term goals, and determinantal point processes to introduce diversity.
Bio
Trevor Potter is a Software Engineer at Google working on the YouTube Recommendations team. Prior to joining YouTube in 2014, he completed a Ph.D. in Applied Mathematics and UC Berkeley.