Event Date
Speaker: Akos Lada, Data Science Manager, Facebook.
Title: "Facebook's News Feed Ranking and Estimating Long-Run Outcomes"
Abstract: Facebook’s News Feed ranking system is a highly personalized machine learning system with content for more than 2 billion people. The ranking system aims to show each user the content that is the most relevant and meaningful for them, every time they come to Facebook. Whenever we make any changes to the algorithm we want to understand if the change will be beneficial to our users in the long term. But to understand the long-term stable effects of an experiment, decision-makers often need to be able to run the experiment over a long horizon, which is often infeasible because of resources and timelines. We present a surrogacy method we developed at Facebook’s News Feed team that helps us make quicker decisions whether an experiment is working well based on short-term outcomes. As a result, we can make quicker decisions about randomized experiments whether they are beneficial to users, and we can use Bayesian optimization to optimize for long-run beneficial outcomes.
About the speaker: Akos Lada is currently a data science manager in Facebook. He leads the News Feed Relevance data science team (which includes News Feed, Stories and Notifications ranking) at Facebook across the Menlo Park and New York offices of Facebook.
Seminar Date/Time: Thursday February 11, 4:10pm
This seminar will be delivered remotely via Zoom. To access the Zoom meeting for this seminar, please contact the instructor Xiucai Ding (xcading@ucdavis.edu) or Pete Scully (pscully@ucdavis.edu) for the meeting ID and password, stating your affiliation.