Research map / Bandits and online learning / Multi-armed bandits
Communication-Constrained Bandits under Additive Gaussian Noise
Prathamesh Mayekar, Jonathan Scarlett, Vincent Tan
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use multi-subphase uniform exploration with successively refined reward centering and scaling | UE-UCB++ with CAS · Distributed stochastic multi-armed bandit over AWGN | regret upper bound: \frac{8 K B \log B}{\mathtt{SNR} \wedge 1}+6KB + 8 \sqrt{\left(\frac{4}{\mathtt{SNR}}+1\right)} \cdot \sqrt{ KT \ln T} | Improved |
| Center each reward using the previous subphase's mean estimate and scale according to its second-moment bound | UE-UCB++ · Distributed stochastic multi-armed bandit over AWGN | variance of mean estimates | Improved |
| Use zero side information rather than sending learned mean estimates (ablation) | UCB0 with CAS relative to UE-UCB++ · Distributed stochastic multi-armed bandit over AWGN | regret upper bound: 8 \sqrt{ \frac{B^2}{\mathtt{SNR}}+1} \cdot \sqrt{KT \ln T} + 6 KB; \frac{8 K B \log B}{\mathtt{SNR} \wedge 1}+6KB + 8 \s | Hurt |
| Use a single uniform-exploration phase instead of the refined multi-subphase scheme (ablation) | UE-UCB relative to UE-UCB++ · Distributed stochastic multi-armed bandit over AWGN | additive regret term: \frac{2KB^3}{\mathtt{SNR}}; \frac{8 K B \log B}{\mathtt{SNR} \wedge 1} | Hurt |
| Replace UCB with MOSS in the final phase (ablation) | UE-UCB++ · Distributed stochastic multi-armed bandit over AWGN | regret upper-bound logarithmic factor: The \ln T factor ... can be removed | Improved |
| Use CAS encoding rather than quantization followed by transmission over AWGN | Distributed bandit communication · Distributed stochastic multi-armed bandit over AWGN | regret | Improved |
Closest papers by meaning
- Batched Stochastic Linear Bandits with 1-Bit Communication Constraints (NeurIPS 2026 · similarity 0.91)
- Individual Regret in Cooperative Stochastic Multi-Armed Bandits (NeurIPS 2025 · similarity 0.90)
- Improved Regret Analysis in Gaussian Process Bandits: Optimality for Noiseless Reward, RKHS norm, and Non-Stationary Variance (ICML 2025 · similarity 0.89)
- Stochastic Bandits Robust to Adversarial Attacks (ICLR 2025 · similarity 0.89)
- Small steps no more: Global convergence of stochastic gradient bandits for arbitrary learning rates (NeurIPS 2024 · similarity 0.89)
- Distributed Multi-Agent Bandits Over Erdős-Rényi Random Networks (NeurIPS 2025 · similarity 0.88)
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.
