Research map / Federated learning and privacy / Personalized and heterogeneous FL
Deploying Models to Non-participating Clients in Federated Learning without Fine-tuning: A Hypernetwork-based Approach
Yuhao Zhou, Jindi Lv, Yuxin Tian, Dan Si, Qing Ye, Jiancheng Lv
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 |
|---|---|---|---|
| generate classifier parameters from embeddings (Opt. 2) instead of conditioning inputs (Opt. 1) (ablation) | HyperFedZero · EMNIST MLP (N=10, alpha_d=1.0) | zACC: 72.13 -> 76.82 | Improved |
| generate classifier parameters from embeddings (Opt. 2) instead of conditioning inputs (Opt. 1) (ablation) | HyperFedZero · FMNIST MLP (N=50, alpha_d=1.0) | zACC: 84.82 -> 91.52 | Improved |
| use Opt. 1 input conditioning instead of plain FedAvg (ablation) | FedAvg · FMNIST MLP (N=50, alpha_d=1.0) | zACC: 86.16 -> 84.82 | Hurt |
| add the balancing penalty variance term (alpha 0 -> 1) (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 5.62 -> 14.68 | Improved |
| increase alpha from 1 to 2 (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 8.75 | Hurt |
| add the entropy term (beta 0 -> 1) (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 8.43 -> 14.68 | Improved |
| increase embedding dimension from 16 to 64 (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 8.43 | Hurt |
| reduce embedding dimension from 16 to 2 (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 3.12 | Hurt |
| shrink hypernetwork hidden sizes from [300,300] to [100,100] (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 7.5 | Hurt |
| enlarge hypernetwork hidden sizes from [300,300] to [500,500] (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 6.87 | Hurt |
| increase chunk size from 576 to 2304 (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 7.81 | Hurt |
| reduce chunk size from 576 to 144 (ablation) | HyperFedZero · Tiny-ImageNet ResNet (N=50, alpha_d=1.0) | zACC: 14.68 -> 7.50 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Scaffold — SCAFFOLD: Stochastic Controlled Averaging for Federated Learning (ICML 2020) · better on 4 of 4 shared results
- FedSR — FedSR: A Simple and Effective Domain Generalization Method for Federated Learning (NeurIPS 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FLUX: Efficient Descriptor-Driven Clustered Federated Learning under Arbitrary Distribution Shifts (NeurIPS 2025 · similarity 0.89)
- A Bayesian Approach for Personalized Federated Learning in Heterogeneous Settings (NeurIPS 2024 · similarity 0.88)
- FedWon: Triumphing Multi-domain Federated Learning Without Normalization (ICLR 2024 · similarity 0.88)
- Bridging Model Heterogeneity in Federated Learning via Uncertainty-based Asymmetrical Reciprocity Learning (ICML 2024 · similarity 0.88)
- FedLMT: Tackling System Heterogeneity of Federated Learning via Low-Rank Model Training with Theoretical Guarantees (ICML 2024 · similarity 0.88)
- Federated Learning with Heterogeneous LLMs: Integrating Small Student Client Models with a Large Hungry Model (AAAI 2025 · similarity 0.87)
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.
