Research map / Learning theory and algorithms / Sample complexity
Distributed Mean Estimation with Limited Communication
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 |
|---|---|---|---|
| Increase the number of quantization levels from 2 to k (ablation) | Stochastic quantization · Distributed mean estimation | MSE: Θ(d/n) -> O(d/(n(k-1)^2)) | Helped |
| Apply a structured random rotation before quantization (ablation) | Stochastic k-level quantization · Distributed mean estimation | MSE: O(d/(n(k-1)^2)) -> O((log d)/(n(k-1)^2)) | Helped |
| Use variable-length coding instead of fixed-length bin codes (ablation) | Stochastic k-level quantization · Distributed mean estimation | communication cost: n · (d ⌈log₂ k⌉ + Õ(1)) -> O(nd(1 + log(k²/d + 1)) + Õ(n)) | Helped |
| Set k to √d + 1 and use variable-length coding (ablation) | Variable-length coded quantization · Distributed mean estimation | MSE: O(d/n) -> O(1/n) | Helped |
| Sample clients with participation probability p (ablation) | Quantization protocols · Distributed mean estimation | communication cost: C(π_p, X^n) = p · C(π, X^n) | Helped |
| Sample clients with participation probability p (ablation) | Quantization protocols · Distributed mean estimation | MSE: E(π_p, X^n) = (1/p) · E(π, X^n) + (1-p)/(np) Σᵢ ||Xᵢ||²₂ | Hurt |
| Combine structured random rotation with variable-length coding (ablation) | Stochastic quantization · Distributed mean estimation | communication cost | No clear effect |
Closest papers by meaning
- New Bounds For Distributed Mean Estimation and Variance Reduction (ICLR 2021 · similarity 0.90)
- Correlated Quantization for Distributed Mean Estimation and Optimization (ICML 2022 · similarity 0.90)
- Optimal Sparsity-Sensitive Bounds for Distributed Mean Estimation (NeurIPS 2019 · similarity 0.89)
- Correlation Aware Sparsified Mean Estimation Using Random Projection (NeurIPS 2023 · similarity 0.89)
- Accelerating Federated Learning with Quick Distributed Mean Estimation (ICML 2024 · similarity 0.87)
- Distributed Nonparametric Estimation: from Sparse to Dense Samples per Terminal (ICML 2025 · 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.