Research map / Federated learning and privacy / Communication-efficient FL
Every Parameter Matters: Ensuring the Convergence of Federated Learning with Dynamic Heterogeneous Models Reduction
Hanhan Zhou, Tian Lan, Guru Prasadh Venkataramani, Wenbo Ding
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 minimum coverage index through optimized client-model distribution rather than greedy pruning | Pruning-Greedy · MNIST IID, high model reduction | accuracy: 95.01 -> 95.32 | Improved |
| Increase minimum coverage index through optimized client-model distribution rather than greedy pruning | Pruning-Greedy · MNIST non-IID L=2, high model reduction | accuracy: 76.64 -> 81.66 | Improved |
| Increase minimum coverage index through optimized client-model distribution rather than greedy pruning | Pruning-Greedy · CIFAR-10 IID, high model reduction | accuracy: 67.35 -> 67.74 | Improved |
| Increase minimum coverage index through optimized client-model distribution rather than greedy pruning | Pruning-Greedy · CIFAR-10 non-IID L=2, high model reduction | accuracy: 22.55 -> 27.97 | Improved |
| Increase minimum coverage index through optimized client-model distribution rather than greedy pruning | Pruning-Greedy · CIFAR100 IID, high model reduction | accuracy: 39.29 -> 40.78 | Improved |
| Increase minimum coverage index through optimized client-model distribution rather than greedy pruning | Pruning-Greedy · CIFAR100 non-IID L=20, high model reduction | accuracy: 25.97 -> 26.63 | Improved |
| Distribute extracted subnet regions across clients instead of keeping only leading subnet parameters | Static Subnet Subtraction · MNIST IID, high model reduction | accuracy: 95.88 -> 94.41 | Hurt |
| Distribute extracted subnet regions across clients instead of keeping only leading subnet parameters | Static Subnet Subtraction · MNIST non-IID L=2, high model reduction | accuracy: 71.64 -> 85.82 | Improved |
| Distribute extracted subnet regions across clients instead of keeping only leading subnet parameters | Static Subnet Subtraction · CIFAR-10 IID, high model reduction | accuracy: 68.78 -> 69.15 | Improved |
| Distribute extracted subnet regions across clients instead of keeping only leading subnet parameters | Static Subnet Subtraction · CIFAR-10 non-IID L=2, high model reduction | accuracy: 30.61 -> 33.46 | Improved |
| Distribute extracted subnet regions across clients instead of keeping only leading subnet parameters | Static Subnet Subtraction · CIFAR100 IID, high model reduction | accuracy: 41.18 -> 37.42 | Hurt |
| Distribute extracted subnet regions across clients instead of keeping only leading subnet parameters | Static Subnet Subtraction · CIFAR100 non-IID L=20, high model reduction | accuracy: 26.23 -> 22.40 | Hurt |
Closest papers by meaning
- An Aggregation-Free Federated Learning for Tackling Data Heterogeneity (CVPR 2024 · similarity 0.89)
- FIARSE: Model-Heterogeneous Federated Learning via Importance-Aware Submodel Extraction (NeurIPS 2024 · similarity 0.89)
- One-Shot Heterogeneous Federated Learning with Local Model-Guided Diffusion Models (ICML 2025 · similarity 0.89)
- Tackling Intertwined Data and Device Heterogeneities in Federated Learning with Unlimited Staleness (AAAI 2025 · similarity 0.89)
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · similarity 0.89)
- Fake It Till Make It: Federated Learning with Consensus-Oriented Generation (ICLR 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.
