AI Paper Atlas

Research map / Probabilistic and scientific modeling / Treatment effects

Optimized Covariance Design for AB Test on Social Network under Interference

Qianyi Chen, Bo Li, LU DENG, Yong Wang

NeurIPS 2023 · Probabilistic and scientific modeling › Treatment effects · approach: Distributional treatment effects · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use optimized covariance design (OCD) instead of complete randomizationcluster randomization with HT estimator · FB-Stanford3 multiplicative model, gamma=2.0MSE: 2.297 -> 1.073Improved
use OCD instead of PSRcluster randomization with HT estimator · FB-Stanford3 linear model, gamma=2.0MSE: 1.496 -> 0.926Improved
use OCD instead of IBR-p at weak interferencecluster randomization with HT estimator · FB-Stanford3 linear model, gamma=0.5MSE: 0.141 -> 0.209Hurt
use OCD instead of PSR at gamma=1.0cluster randomization with HT estimator · FB-Stanford3 linear model, gamma=1.0MSE: 0.415 -> 0.371Improved
use OCD instead of PSRcluster randomization with HT estimator · FB-Stanford3 multiplicative model, gamma=0.5MSE: 0.152 -> 0.069Improved
use OCD instead of complete randomizationcluster randomization with HT estimator · FB-Stanford3 linear model, gamma=2.0SD: 0.562 -> 0.585No clear effect
use OCD instead of other designs under model misspecification (multiplicative model)cluster randomization with HT estimator · FB-Stanford3 multiplicative model, gamma=1.0MSE: 0.569 -> 0.271Improved
vary Louvain clustering resolution (2, 5, 10) (ablation)OCD · FB-Stanford3 and FB-Cornell5 appendixMSENo clear effect
vary omega in the MSE bound (ablation)OCD · FB-Stanford3MSENo clear effect
use weaker spectral-norm bound on interference term (ablation)OCD objective · FB-Stanford3MSEHurt
use worst-case variance objective (IBR) instead of OCD boundblock randomization · FB-Stanford3MSE: 0.926 -> 1.541Hurt
use pairing blocks of size 2 (IBR-p) instead of IBRindependent block randomization · FB-Stanford3 linear model, gamma=0.5MSE: 0.164 -> 0.141Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy