Research map / Federated learning and privacy / Differential privacy
Federated Linear Contextual Bandits with User-level Differential Privacy
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 ROBIN's phased LinUCB initialization, greedy exploitation, and privatized estimate aggregation | Federated linear contextual bandits · User-level CDP with margin and diversity conditions | regret | Helped |
| Use user-level CDP rather than no privacy constraint | Federated linear contextual bandits · ε=Ω(1/√M), with or without the margin condition | regret | No clear effect |
| Use user-level CDP rather than no privacy constraint | Federated linear contextual bandits · ε=O(1/√M), with margin condition | regret | Hurt |
| Use user-level CDP rather than no privacy constraint | Federated linear contextual bandits · ε=O(1/√M), without margin condition | regret | Hurt |
| Use user-level LDP rather than no privacy constraint | Federated linear contextual bandits · With margin condition | regret | Hurt |
| Use user-level LDP rather than no privacy constraint | Federated linear contextual bandits · Without margin condition | regret | Hurt |
| Use pure user-level LDP instead of approximate user-level LDP | Federated linear contextual bandits · With or without margin condition | regret lower bound: replacing ε in the CDP lower bound by ε/√M | Hurt |
| Use independent local arm-pulling decisions without information sharing | Federated linear contextual bandits under user-level LDP · ε=O(1/M) | regret | No clear effect |
| Aggregate local estimates with user-level CDP | Federated linear contextual bandits · ROBIN's phased estimates | regret: \tilde{O}\left(\max\left(1, \frac{d\log T\log(\frac{1}{\delta})\log^3(\frac{1}{\beta})}{M\varepsilon^2 }\right)\frac{C_0 | Helped |
| Require the margin condition in addition to diversity | Federated linear contextual bandits · ROBIN's regret analysis | regret: ROBIN has a regret upper bound; without the margin condition, the stated lower bound is \Omega\left( \min\left\{M, \max\ | Helped |
Closest papers by meaning
- On Differentially Private Federated Linear Contextual Bandits (ICLR 2024 · similarity 0.93)
- Shuffle and Joint Differential Privacy for Generalized Linear Contextual Bandits (NeurIPS 2026 · similarity 0.91)
- Federated Contextual Cascading Bandits with Asynchronous Communication and Heterogeneous Users (AAAI 2024 · similarity 0.88)
- Communication-Efficient Federated Non-Linear Bandit Optimization (ICLR 2024 · similarity 0.88)
- Federated X-armed Bandit (AAAI 2024 · similarity 0.87)
- Federated Online Prediction from Experts with Differential Privacy: Separations and Regret Speed-ups (NeurIPS 2024 · similarity 0.87)
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.