Research map / Bandits and online learning / Multi-armed bandits
Federated Multi-Armed Bandits
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 |
|---|---|---|---|
| Gradually admit new clients while accounting for arm-sampling and client-sampling uncertainty | Fed2-UCB · Approximate FMAB | regret: O(log(T)) | Helped |
| Use Fed1-UCB with a logarithmic update period | Fed1-UCB · Exact FMAB | regret: O(∑_{k≠k_*} σ²log(T)/Δ_k) | Helped |
| Use only M=200 clients (ablation) | Fed1-UCB · MovieLens | regret | Hurt |
| Involve all clients rather than sampling only as needed (ablation) | Fed1-UCB · MovieLens | regret | Hurt |
| Gradually sample clients with f(p)=200 and g(p)=2^p | Fed2-UCB · MovieLens | regret | Helped |
| Set f(p)=500, admitting clients more slowly (ablation) | Fed2-UCB · MovieLens | regret | Helped |
| Replace logarithmic update periods with raw-data communication every time slot (ablation) | Fed1-UCB · Synthetic cognitive radio, exact model | regret | Hurt |
| Increase the fixed client count from M=5 to M=10 (ablation) | Fed1-UCB · Synthetic cognitive radio, exact model | regret | No clear effect |
| Set Fed2-UCB f(p)=100 and g(p)=2^p | Fed2-UCB · Synthetic cognitive radio, approximate model | regret | Mixed |
| Use f(p)=10, admitting clients rapidly (ablation) | Fed2-UCB · Synthetic cognitive radio, approximate model, T=10^4 | regret | Hurt |
| Use f(p)=100, ensuring more local exploration (ablation) | Fed2-UCB · Synthetic cognitive radio, approximate model, T=10^4 | regret | Hurt |
| Balance client admission and local exploration with f(p)=50 (ablation) | Fed2-UCB · Synthetic cognitive radio, approximate model, T=10^4 | regret | Helped |
Closest papers by meaning
- Communication-Efficient Federated Non-Linear Bandit Optimization (ICLR 2024 · similarity 0.90)
- Federated X-armed Bandit (AAAI 2024 · similarity 0.89)
- Federated Combinatorial Multi-Agent Multi-Armed Bandits (ICML 2024 · similarity 0.89)
- Incentivized Truthful Communication for Federated Bandits (ICLR 2024 · similarity 0.88)
- Federated Contextual Cascading Bandits with Asynchronous Communication and Heterogeneous Users (AAAI 2024 · similarity 0.88)
- Federated Q-Learning: Linear Regret Speedup with Low Communication Cost (ICLR 2024 · similarity 0.86)
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.