AI Paper Atlas

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

An Empirical Process Approach to the Union Bound: Practical Algorithms for Combinatorial and Linear Bandits

Julian Katz-Samuels, Lalit Jain, zohar karnin, Kevin G. Jamieson

NeurIPS 2020 · Bandits and online learning › Multi-armed bandits · approach: Pure exploration algorithms · 24 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 FBPeace instead of uniform samplingFixed-budget multivariate testing · Multivariate testing (k=6, d=3)likelihood of returning the optimal layout: 30% more likely at 10000 samplesImproved
Use Peace rather than RAGE as the number of candidate arms increasesLinear bandit identification · Linear bandits (10^3 to 10^6 arms)sample count: constant vs. linearly in log(|Z|)Improved
Use Peace rather than CLUCB to identify a close second-best matchingCombinatorial bandit identification · Matchingsample complexity as the reward gap h shrinks: roughly |U|/h^2 -> roughly 1/h^2Improved
Use Peace rather than CLUCB for shortest-path identificationCombinatorial bandit identification · Shortest Pathsample complexity as the reward gap h shrinksImproved
Optimize the experimental design for Gaussian width instead of relying on a union bound over candidate armsPeace · Transductive linear banditssample complexity dependence on the candidate set: avoids an explicit union bound over the number of armsImproved

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