AI Paper Atlas

Research map / Bandits and online learning / Contextual bandits and Thompson sampling

Regret Bounds for Gaussian-Process Optimization in Large Domains

Manuel Wuthrich, Bernhard Schölkopf, Andreas Krause

NeurIPS 2021 · Bandits and online learning › Contextual bandits and Thompson sampling · approach: Gaussian process bandit regret · 3 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 EI (always maximise) instead of EI2 (alternate max/min evaluations) (ablation)EI · continuous GP samples, 1000 grid pointsevaluations to reach regret: 0.5t_ei2 <= t_ei <= t_ei2Improved
Use EI instead of EI2 (ablation)EI · continuous D=3, l=0.030, R=1.5step ratio: 1.06Hurt
Use EI instead of EI2 (ablation)EI · continuous D=4, l=0.300, R=0.5step ratio: 0.70Improved
Use UCB instead of UCB2 (ablation)UCB · continuous domains with fixed confidence levelstep ratio: 0.66Improved
Use UCB instead of UCB2 (ablation)UCB · continuous D=2, l=0.030, R=2.5step ratio: 0.67Improved
Use EI instead of EI2 (ablation)EI · band covariance N=100step ratio: 0.75Improved
Use UCB instead of UCB2 (ablation)UCB · band covariance N=100, band_size=40step ratio: 1.00No clear effect
Use EI instead of EI2 (ablation)EI · Wishart covariance N=200step ratio: 0.67Improved
Use UCB instead of UCB2 (ablation)UCB · Wishart covariance N=200step ratio: 0.81Improved
Increase length-scale l to large values with low dimensionEI vs EI2 · continuous domainsstep ratio: 1.00No clear effect
Very small length-scale (no correlation between points)EI vs EI2 · continuous domainsstep ratio: 1.00No clear effect

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