Research map / Bandits and online learning / Online convex optimization and games
Nonstochastic Multiarmed Bandits with Unrestricted Delays
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use DEW for variable bounded delays with a known delay upper bound | Delayed Exp3 · Nonstochastic bandits, variable bounded delays | expected regret bound: 2 \sqrt{ \left ( \frac{KTe}{2} + D \right) \ln K} | Helped |
| Skip rounds whose delays meet or exceed beta | DEW · Nonstochastic bandits, unrestricted delays | expected regret bound: 2\sqrt{ \left ( \frac{KTe}{2} + (1 + 4e)D \right) \ln K} | Helped |
| Tune the skipping threshold by doubling when delays are available at action time | Skipper + DEW · Nonstochastic bandits, unrestricted delays available at action time | expected regret bound: 15 \min_{\beta} \left \{ \vert S_{\beta} \vert + 4e\beta \ln K + \frac{KT + D_{\beta}}{4e\beta} \right \} + 10 e^{2} K \ | Helped |
| Use the oracle bound rather than the explicit bound when D_beta is much smaller than D | Doubling + Skipper + DEW · Example with a large delay on early rounds and zero delay on later rounds | regret bound: explicit: \mathcal{O} \left (\sqrt{KT\ln K + T\sqrt{KT}} \right ) = \mathcal{O}\big(T^{3/4}\big); oracle: \mathcal{O} \l | Helped |
Closest papers by meaning
- Near-Optimal Stochastic Linear Bandits with Delay (NeurIPS 2026 · similarity 0.90)
- Delay as Payoff in MAB (AAAI 2025 · similarity 0.90)
- Regret Bounds for Adversarial Contextual Bandits with General Function Approximation and Delayed Feedback (NeurIPS 2025 · similarity 0.89)
- Contextual Linear Bandits with Delay as Payoff (ICML 2025 · similarity 0.89)
- A Best-of-both-worlds Algorithm for Bandits with Delayed Feedback with Robustness to Excessive Delays (NeurIPS 2024 · similarity 0.89)
- Online Nonsubmodular Optimization with Delayed Feedback in the Bandit Setting (AAAI 2025 · similarity 0.89)
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.