Research map / Optimization / Distributed and compressed
Correlated Quantization for Distributed Mean Estimation and Optimization
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 correlated rather than independent stochastic quantization | Distributed mean estimation · Synthetic | mean squared error: 10.28 (0.25) -> 1.40 (0.05) | Helped |
| Use correlated rather than independent stochastic quantization | Distributed mean estimation · MNIST | mean squared error: 0.466 (0.014) -> 0.141 (0.004) | Helped |
| Add random rotation to correlated quantization (ablation) | Correlated quantization · Synthetic distributed mean estimation | mean squared error: 1.40 (0.05) -> 1.01 (0.06) | Helped |
| Add random rotation to correlated quantization (ablation) | Correlated quantization · MNIST distributed mean estimation | mean squared error: 0.141 (0.004) -> 0.238 (0.012) | Hurt |
| Add random rotation to correlated quantization (ablation) | Correlated quantization · MNIST distributed k-means | objective: 39.97 (0.13) -> 42.21 (0.17) | Hurt |
| Add random rotation to correlated quantization (ablation) | Correlated quantization · MNIST distributed power iteration | error: 0.055 (0.002) -> 0.059 (0.002) | Hurt |
| Add random rotation to correlated quantization (ablation) | Correlated quantization · Federated MNIST FedAvg | accuracy: 88.39 (0.12) -> 89.20 (0.04) | Helped |
| Use correlated quantization instead of independent quantization | Distributed k-means · MNIST | objective: 42.98 (0.23) -> 39.97 (0.13) | Helped |
| Use correlated quantization instead of independent quantization | Distributed power iteration · MNIST | error: 0.242 (0.005) -> 0.055 (0.002) | Helped |
| Use correlated quantization instead of independent quantization | FedAvg · Federated MNIST | accuracy: 87.38 (0.30) -> 88.39 (0.12) | Helped |
| Use shared randomness to correlate clients' quantization | Stochastic quantization · Mean estimation when inputs are concentrated | mean squared error | Helped |
| Apply random rotation | Correlated and independent quantization · Synthetic mean estimation experiment varying client count | mean squared error | Helped |
Closest papers by meaning
- Distributed Mean Estimation with Limited Communication (ICML 2017 · similarity 0.90)
- Correlation Aware Sparsified Mean Estimation Using Random Projection (NeurIPS 2023 · similarity 0.89)
- New Bounds For Distributed Mean Estimation and Variance Reduction (ICLR 2021 · similarity 0.88)
- Distributed Nonparametric Estimation: from Sparse to Dense Samples per Terminal (ICML 2025 · similarity 0.85)
- Accelerating Federated Learning with Quick Distributed Mean Estimation (ICML 2024 · similarity 0.85)
- JointSQ: Joint Sparsification-Quantization for Distributed Learning (CVPR 2024 · similarity 0.84)
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.