Modeling interference using experiment roll-out

A Boyarsky, H Namkoong, J Pouget-Abadie - arxiv preprint arxiv …, 2023 - arxiv.org
Experiments on online marketplaces and social networks suffer from interference, where the
outcome of a unit is impacted by the treatment status of other units. We propose a framework …

On online experimentation without device identifiers

S Shankar, R Sinha, M Fiterau - Forty-first International Conference …, 2024 - openreview.net
Measuring human feedback via randomized experimentation is a cornerstone of data-driven
decision-making. The methodology used to estimate user preferences from their online …

Experimental design for causal inference through an optimization lens

J Zhao - … in Operations Research: Smarter Decisions for a …, 2024 - pubsonline.informs.org
The study of experimental design offers tremendous benefits for answering causal questions
across a wide range of applications, including agricultural experiments, clinical trials …

A/B testing under Interference with Partial Network Information

S Shankar, R Sinha, Y Chandak… - International …, 2024 - proceedings.mlr.press
A/B tests are often required to be conducted on subjects that might have social connections.
For eg, experiments on social media, or medical and social interventions to control the …

Causal Inference in Social Platforms Under Approximate Interference Networks

Y Jiang, L Deng, Y Wang, H Wang - arxiv preprint arxiv:2408.04441, 2024 - arxiv.org
Estimating the total treatment effect (TTE) of a new feature in social platforms is crucial for
understanding its impact on user behavior. However, the presence of network interference …

Cascade-based Randomization for Inferring Causal Effects under Diffusion Interference

Z Fatemi, J Pouget-Abadie, E Zheleva - Proceedings of the International …, 2024 - ojs.aaai.org
The presence of interference, where the outcome of an individual may depend on the
treatment assignment and behavior of neighboring nodes, can lead to biased causal effect …

Causal Inference under Network Interference Using a Mixture of Randomized Experiments

Y Jiang, H Wang - arxiv preprint arxiv:2309.00141, 2023 - arxiv.org
In randomized experiments, the classic stable unit treatment value assumption (SUTVA)
states that the outcome for one experimental unit does not depend on the treatment …

Country-diverted experiments for mitigation of network effects

L Lin, C Meng, J Brennan, J Pouget-Abadie… - Proceedings of the 18th …, 2024 - dl.acm.org
For large content platforms, user-diverted A/B experimentation has become the standard
way of evaluating impact of recommendation system changes on audience engagement and …

The Conflict Graph Design: Estimating Causal Effects under Arbitrary Neighborhood Interference

V Kandiros, C Pipis, C Daskalakis… - arxiv preprint arxiv …, 2024 - arxiv.org
A fundamental problem in network experiments is selecting an appropriate experimental
design in order to precisely estimate a given causal effect of interest. In fact, optimal rates of …

A Two-Part Machine Learning Approach to Characterizing Network Interference in A/B Testing

Y Yuan, KM Altenburger - arxiv preprint arxiv:2308.09790, 2023 - arxiv.org
The reliability of controlled experiments, or" A/B tests," can often be compromised due to the
phenomenon of network interference, wherein the outcome for one unit is influenced by …