Research map / Federated learning and privacy / Communication-efficient FL
EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning
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 EDEN's 2-bit quantization instead of its 1-bit quantization (ablation) | EDEN · Standard normal quantization, d→∞ | vNMSE: 0.571 -> 0.134 | Helped |
| Assign coordinates randomly between adjacent quantization budgets rather than splitting the vector into halves (ablation) | EDEN at 1.5 bits per coordinate · Standard normal quantization, d→∞ | vNMSE: 0.352 -> 0.317 | Helped |
| Add random sparsification to 1-bit quantization for sub-bit compression (ablation) | EDEN · Standard normal quantization, 0.1 bits per coordinate, d→∞ | vNMSE: approximately 14.707 | Hurt |
| Use entropy-constrained intervals instead of I₃ without entropy encoding (ablation) | EDEN · Standard normal quantization, 3 bits per coordinate | vNMSE: 0.03572 -> 0.022741 | Helped |
| Use entropy-constrained intervals instead of a 9-value Lloyd-Max quantizer with entropy encoding (ablation) | EDEN · Standard normal quantization, 3 bits per coordinate | vNMSE: improvement of approximately 20% | Helped |
| Use a structured Hadamard rotation instead of a fully random rotation (ablation) | EDEN · Evaluation experiments | vNMSE | No clear effect |
| Allow packet loss and replace missing coordinates with zeros, rescaling the received vector (ablation) | EDEN · Lossy networks with constant packet loss rate | vNMSE | Hurt |
| Use EDEN's entropy-constrained intervals rather than the Lloyd-Max 9-value quantizer (ablation) | EDEN · Standard normal quantization, 3 bits per coordinate | vNMSE: nearly 20% | Helped |
Closest papers by meaning
- Accelerating Federated Learning with Quick Distributed Mean Estimation (ICML 2024 · similarity 0.93)
- Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees (ICLR 2025 · similarity 0.88)
- Communication-Efficient Federated Learning with Accelerated Client Gradient (CVPR 2024 · similarity 0.88)
- FedSMU: Communication-Efficient and Generalization-Enhanced Federated Learning through Symbolic Model Updates (ICML 2025 · similarity 0.87)
- FedAdaVR: Adaptive Variance Reduction for Robust Federated Learning under Limited Client Participation (NeurIPS 2026 · similarity 0.86)
- Low Precision Local Training is Enough for Federated Learning (NeurIPS 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.