Back to the program

Invited Paper Session

Integrating Algorithms and Analysis for Adaptively Randomized Experiments

Sun, Aug 6, 4:00 PM

Joseph Jay WilliamsOrganizerAaditya RamdasOrganizerJohn LangfordChairTong LiOrganizer

Section on Statistical Learning and Data Science

6 Presentations

A Bandit Approach to Multiple Testing with False Discovery Control

4:05 PM - 4:20 PM

Kevin Jamieson (University of Washington)

A review of adaptive experiments conducted at Meta

4:20 PM - 4:35 PM

Eytan Bakshy (Facebook)

Algorithms for Adaptive Experiments that Trade-off Statistical Analysis with Reward: Adaptively Combining Uniform Random Assignment and Thompson Sampling

4:35 PM - 4:50 PM

Co-authors: Joseph Jay Williams (University of Toronto Statistics, Computer Science)

Anytime-valid off-policy inference for contextual bandits

4:50 PM - 5:05 PM

Ian Waudby-Smith (Carnegie Mellon University)

Efficient Inference Without Trading-off Regret in Bandits: An Allocation Probability Test for Thompson Sampling

5:05 PM - 5:20 PM

Nina Deliu (Sapienza University of Rome)

Statistical Inference After Using Online Reinforcement Learning on Longitudinal Data

5:20 PM - 5:35 PM

Discussant

Sofia Villar (University of Cambridge)

↑