AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

Experimental Designs for Heteroskedastic Variance

Justin David Naggar Weltz, Tanner Fiez, Alexander Volfovsky, Eric Laber, Blake Mason, houssam nassif, Lalit K Jain

NeurIPS 2023 · Learning theory and algorithms › Sample complexity · approach: Coresets and sublinear estimation · 0 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
HEAD adaptive-design variance estimator versus Separate Arm and Uniform estimators (ablation)HEAD · synthetic arm sets X1 (200 unit-sphere arms) and X2 (1800 radius-0.1 arms), d=15mean absolute error of variance estimatesImproved
HEAD versus competing estimators across dimensions at fixed sample size 95k (ablation)HEAD · synthetic arm sets, varying dmean absolute error of variance estimatesImproved
estimate heteroskedastic variances (H-RAGE) instead of assuming homoskedastic noise (RAGE) (ablation)H-RAGE · Example 1 signal-to-noise linear bandit, d=4, omega=0.02, q=0.4sample complexityImproved
estimate heteroskedastic variances (H-RAGE) instead of RAGE (ablation)H-RAGE · Example 2 benchmark with bent informative arms, d=3, alpha^2=1, beta^2=0.2sample complexityImproved
account for heteroskedastic variances instead of upper bounding noise by sigma_max (ablation)H-RAGE and heteroskedastic oracle · Experiment 3 multivariate testing, 2 variations in 3 dimensionssample complexityImproved
heteroskedastic oracle allocation versus homoskedastic oracle (ablation)Oracle allocation · Experiment 3 multivariate testingsampling weight allocationImproved
increase kappa, the variance ratio, between heteroskedastic and homoskedastic algorithmsOptimal sample complexity comparison · theoretical comparison with kappa=20gap in optimal sample complexityMixed

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