Research map / Federated learning and privacy / Communication-efficient FL
FedDM: Iterative Distribution Matching for Communication-Efficient 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 |
|---|---|---|---|
| replace client model updates with locally learned synthetic data and aggregate their surrogate objectives | federated learning · CIFAR10, Dir10(0.5), 20 communication rounds | test accuracy: 66.12± 0.17% -> 69.66± 0.13% | Helped |
| use iterative distribution matching to send synthetic data rather than model updates | federated learning · CIFAR10, Dir10(0.1) | test accuracy: 60.96 ± 1.20 -> 67.38 ± 0.32 | Helped |
| use iterative distribution matching to send synthetic data rather than model updates | federated learning · CIFAR10, Dir10(0.01) | test accuracy: 57.32 ± 0.04 -> 63.82 ± 0.17 | Helped |
| transmit learned synthetic images rather than real images of the same size | FedDM · CIFAR10, Dir10(0.5) | test accuracy: 68.66 ± 0.08% -> 69.62 ± 0.14% | Helped |
| increase synthetic images per class from 3 to 5 (ablation) | FedDM · CIFAR10, Dir10(0.5) | test accuracy: 53.64 ± 0.35% -> 62.24 ± 0.04% | Helped |
| increase synthetic images per class from 5 to 10 (ablation) | FedDM · CIFAR10, Dir10(0.5) | test accuracy: 62.24 ± 0.04% -> 69.62 ± 0.14% | Helped |
| vary the surrogate radius rho among 3, 5, and 10 (ablation) | FedDM · CIFAR10, Dir10(0.5) | test accuracy: 69.15 ± 0.09% (rho=3); 69.66 ± 0.13% (rho=5); 69.32 ± 0.24% (rho=10) | No clear effect |
| increase communication payload by increasing synthetic images per class (ablation) | FedDM · CIFAR10, Dir10(0.5) | communication cost | Hurt |
Closest papers by meaning
- FLUX: Efficient Descriptor-Driven Clustered Federated Learning under Arbitrary Distribution Shifts (NeurIPS 2025 · similarity 0.88)
- Progressive Distribution Matching for Federated Semi-Supervised Learning (AAAI 2025 · similarity 0.88)
- Data-Free Black-Box Federated Learning via Zeroth-Order Gradient Estimation (AAAI 2025 · similarity 0.88)
- FedWon: Triumphing Multi-domain Federated Learning Without Normalization (ICLR 2024 · similarity 0.88)
- Federated Learning via Input-Output Collaborative Distillation (AAAI 2024 · similarity 0.88)
- FedLMT: Tackling System Heterogeneity of Federated Learning via Low-Rank Model Training with Theoretical Guarantees (ICML 2024 · similarity 0.88)
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.