Research map / Federated learning and privacy / Personalized and heterogeneous FL
SemiFL: Semi-Supervised Federated Learning for Unlabeled Clients with Alternate Training
Enmao Diao, Jie Ding, Vahid Tarokh
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 |
|---|---|---|---|
| fine-tune the aggregated global model with labeled server data each round (ablation) | SemiFL · CIFAR10 4000 labels, IID | accuracy: 47.03 -> 93.10 | Improved |
| generate pseudo-labels once with the received global model (ablation) | SemiFL · CIFAR10 4000 labels, IID | accuracy: 81.70 -> 93.10 | Improved |
| fine-tune global model with labeled data (ablation) | SemiFL · CIFAR10 4000 labels, Non-IID K=2 | accuracy: 48.89 -> 85.34 | Improved |
| generate pseudo-labels with global model (ablation) | SemiFL · CIFAR10 4000 labels, Non-IID K=2 | accuracy: 80.42 -> 85.34 | Improved |
| alternate training (both components) over vanilla FedAvg + FixMatch (ablation) | FedAvg + FixMatch · CIFAR10 4000 labels, IID | accuracy: 40.26 -> 93.10 | Improved |
| use 5 local epochs instead of 1 (ablation) | SemiFL · CIFAR10 4000 labels, IID | accuracy: 88.9(0.3) -> 93.1(0.1) | Improved |
| add Mixup data augmentation (ablation) | SemiFL · CIFAR10 4000 labels, IID | accuracy: 91.3(0.2) -> 93.1(0.1) | Improved |
| add global momentum 0.5 (ablation) | SemiFL · CIFAR10 4000 labels, IID | accuracy: 92.4(0.1) -> 93.1(0.1) | Improved |
| add global momentum 0.5 (ablation) | SemiFL · CIFAR10 4000 labels, Non-IID K=2 | accuracy: 85.4(0.6) -> 85.3(0.3) | No clear effect |
| update sBN statistics with server and clients instead of server only (ablation) | SemiFL · CIFAR10 4000 labels, IID | accuracy: 93.1(0.2) -> 93.1(0.1) | No clear effect |
| SemiFL instead of existing SSFL methods (FedMatch) | FedMatch · CIFAR10 4000 labels, IID | accuracy: 58.6(0.5) -> 93.1(0.1) | Improved |
Later papers that compared with this method
2 later accepted papers report SemiFL as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- BSemiFL: Semi-supervised Federated Learning via a Bayesian Approach (ICML 2025 · better on 12 of 13 shared results)
- (FL)$^2$: Overcoming Few Labels in Federated Semi-Supervised Learning (NeurIPS 2024 · better on 4 of 4 shared results)
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
- (FL)$^2$: Overcoming Few Labels in Federated Semi-Supervised Learning (NeurIPS 2024 · similarity 0.94)
- SemiDFL: A Semi-Supervised Paradigm for Decentralized Federated Learning (AAAI 2025 · similarity 0.93)
- BSemiFL: Semi-supervised Federated Learning via a Bayesian Approach (ICML 2025 · similarity 0.92)
- Exploring One-Shot Semi-supervised Federated Learning with Pre-trained Diffusion Models (AAAI 2024 · similarity 0.91)
- Progressive Distribution Matching for Federated Semi-Supervised Learning (AAAI 2025 · similarity 0.91)
- Combating Data Imbalances in Federated Semi-supervised Learning with Dual Regulators (AAAI 2024 · similarity 0.91)
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.
