Research map / Federated learning and privacy / Personalized and heterogeneous FL
Federated semi-supervised and graph learning
Learn from unlabeled or graph data across clients with pseudo-labeling.
84 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2020–2026) · an approach in Personalized and heterogeneous FL, keyword cluster “domain · label · multi”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
20 papers in 2023–24 and 54 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 8.0% to 13.9%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2020 · Federated Learning with Only Positive Labels (ICML 2020 · 42 citations)
- 2021 · Federated Semi-Supervised Learning with Inter-Client Consistency & Disjoint Learning (ICLR 2021 · 23 citations)
- 2022 · RSCFed: Random Sampling Consensus Federated Semi-Supervised Learning (CVPR 2022 · 101 citations)
- 2023 · Class Balanced Adaptive Pseudo Labeling for Federated Semi-Supervised Learning (CVPR 2023 · 64 citations)
- 2024 · Federated Graph Learning under Domain Shift with Generalizable Prototypes (AAAI 2024 · 46 citations)
- 2025 · Exploring Vacant Classes in Label-Skewed Federated Learning (AAAI 2025 · 21 citations)
- 2026 · Causally-Aware Attribute Completion for Incomplete Federated Graph Clustering (AAAI 2026 · 3 citations)
Earliest papers here
- Federated Learning with Only Positive Labels (ICML 2020 · 42 citations)
- Federated Semi-Supervised Learning with Inter-Client Consistency & Disjoint Learning (ICLR 2021 · 23 citations)
- RSCFed: Random Sampling Consensus Federated Semi-Supervised Learning (CVPR 2022 · 101 citations)
What has been tried
52 papers here report 547 measured changes (a component added, removed or swapped, compared with a baseline): helped 77%, no clear effect 5%, hurt 15%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| add FCPP | Twitch α=10 (AVG) · accuracy | 50.04 -> 61.74 | Helped | Federated Graph Learning under Domain Shift with Generalizable Prototypes (AAAI 2024 · 46 citations) |
| use federated filtering instead of current-round-only filter aggregation | CIFAR-10 non-IID, p=0.3, αDir=10, ρ=0.6, τ=0.5 · test accuracy | 83.22±2.61 -> 85.31±2.28 | Helped | FedDiv: Collaborative Noise Filtering for Federated Learning with Noisy Labels (AAAI 2024 · 38 citations) |
| add the Distance Regularizer | CIFAR-10 IID ρ=0.5, τ=0.3 · accuracy | 84.45 -> 87.06 | Helped | FedFixer: Mitigating Heterogeneous Label Noise in Federated Learning (AAAI 2024 · 30 citations) |
| Increase the number of cluster centroids from L=3 to L=10 | OpenImage client0 · accuracy | 56.33 -> 57.16 | Helped | Exploring One-Shot Semi-supervised Federated Learning with Pre-trained Diffusion Models (AAAI 2024 · 28 citations) |
| use clustering before concatenating client encoders instead of concatenating all client models | CIFAR-10 · test accuracy | Helped | Exploiting Label Skews in Federated Learning with Model Concatenation (AAAI 2024 · 27 citations) | |
| Add vacant-class distillation and logit suppression to the FedLC objective | CIFAR10, Dirichlet β=0.05 · accuracy | 57.31 -> 75.71 (+18.40) | Helped | Exploring Vacant Classes in Label-Skewed Federated Learning (AAAI 2025 · 21 citations) |
| Add C-reg to the variant without C-reg | CIFAR-10 (DIR, DIR) · accuracy | 54.79 -> 57.73 | Helped | Combating Data Imbalances in Federated Semi-supervised Learning with Dual Regulators (AAAI 2024 · 17 citations) |
| Add local GMM contrastive learning to FedAvg + GCD | CUB-200 and Oxford-Pet EH · All accuracy | 43.3 -> 52.2; 72.1 -> 79.5 | Helped | Federated Generalized Category Discovery (CVPR 2024 · 17 citations) |
| add FUR | CIFAR10 cross-silo, α = 0.1 · KNN accuracy (%) | 80.93 -> 81.39 | Helped | Rethinking the Representation in Federated Unsupervised Learning with Non-IID Data (CVPR 2024 · 17 citations) |
| Add consensus prototype optimizing (CPO) to the baseline client model | All datasets shown in Figure 3 · Clustering performance | Helped | Federated Graph-Level Clustering Network (AAAI 2025 · 10 citations) | |
| Add the weighted aggregation strategy | Child, Insurance, Alarm, Pigs, and Gene · F1 score | Helped | FedCSL: A Scalable and Accurate Approach to Federated Causal Structure Learning (AAAI 2024 · 10 citations) | |
| Add candidate label disambiguation | CIFAR-10, SVHN, CIFAR-100 · accuracy | 53.86% -> 90.35%; 73.77% -> 95.81%; 26.62% -> 68.53% | Helped | Federated Partial Label Learning with Local-Adaptive Augmentation and Regularization (AAAI 2024 · 9 citations) |
What reviewers will expect
Datasets and baselines named in the 27 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CIFAR-10 · 7 of 27 papers
- Cora · 7 of 27 papers
- PubMed · 6 of 27 papers
- CIFAR-100 · 5 of 27 papers
- CiteSeer · 5 of 27 papers
- Fashion-MNIST · 4 of 27 papers
- Amazon Photo · 4 of 27 papers
- MNIST · 3 of 27 papers
Baselines
- FedAvg · 11 of 27 papers
- FedProx · 7 of 27 papers
- FedSage+ · 7 of 27 papers
- FedPub · 7 of 27 papers
- FedStar · 5 of 27 papers
- FedGTA · 4 of 27 papers
- GCFL · 3 of 27 papers
- FedNova · 3 of 27 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 74 papers; from OpenAlex and the papers' affiliations).
- Wuhan University · 5 papers
- The Hong Kong Polytechnic University · 3 papers
- Zhejiang University · 3 papers
- Shanghai Jiao Tong University · 3 papers
- Guangdong University of Technology · 3 papers
- Nanyang Technological University · 3 papers
- Xiamen University · 2 papers
- Northwestern Polytechnical University · 2 papers
- National University of Singapore · 2 papers
- Chinese Academy of Sciences · 2 papers
Use it
Download BibTeX of 30 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Non-IID federated learning (198 papers)
- Federated learning with pretrained models (164 papers)
- Federated prompt and LoRA tuning (91 papers)
- Federated medical and multimodal learning (44 papers)
- Vertical federated learning (38 papers)
Elsewhere in the topic
- Personalization with global-local trade-off (48 papers)
- Sparse and collaboration-graph personalization (36 papers)
- Bayesian personalized FL (35 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).