AI Paper Atlas

Research map / Bandits and online learning / Multi-armed bandits

Box Thirding: Anytime Best Arm Identification under Insufficient Sampling

SEOHWA HWANG, Junyong Park

ICML 2026 · Bandits and online learning › Multi-armed bandits · approach: Fixed-confidence versus fixed-budget identification · 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
use B3 (anytime) instead of BSHanytime BAI · N=128, T=896average simple regret: 0.0992 -> 0.0490Improved
use B3 (anytime) instead of BUCBanytime BAI · N=128, T=896average simple regret: 0.1252 -> 0.0490Improved
use B3 (anytime) instead of BSHanytime BAI · N=1,024, T=10,240average simple regret: 0.1531 -> 0.0268Improved
use B3 (anytime) instead of BUCBanytime BAI · N=1,024, T=10,240average simple regret: 0.2558 -> 0.0268Improved
use B3 (no knowledge of T) instead of UCB-E (knows T)fixed-budget BAI · N=128, T=896average simple regret: 0.0664 -> 0.0490Improved
use B3 (no knowledge of T) instead of UCB-E (knows T)fixed-budget BAI · N=1,024, T=10,240average simple regret: 0.0403 -> 0.0268Improved
use B3 instead of SH with r0=1.7 (which knows T)fixed-budget BAI · N=1,024, T=10,240average simple regret: 0.0195 -> 0.0268Hurt
use B3 instead of SH with r0=1.7 (which knows T)fixed-budget BAI · N=128, T=896average simple regret: 0.0446 -> 0.0490Hurt
change SH base sampling parameter r0 from 2.0 to 1.7SH · N=128, T=896average simple regret: 0.0516 -> 0.0446Improved
change SH base sampling parameter r0 from 2.0 to 1.7SH · N=1,024, T=10,240average simple regret: 0.0288 -> 0.0195Improved
increase reward noise from deterministic to high-noiseUS vs B3 · NYCCC 893, T=10,000average simple regretMixed
replace Gaussian reward noise with Bernoulli or power-law rewardsB3 · NYCCC 893average simple regretNo 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