Research map / Federated learning and privacy / Personalized and heterogeneous FL
Personalization with global-local trade-off
Balance global and personalized federated models with sharpness, prototypes and Gaussian processes.
48 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2021–2026) · an approach in Personalized and heterogeneous FL, keyword cluster “personalized federated · pfl · personalization”
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.
22 papers in 2023–24 and 20 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 8.8% to 5.1%.
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
- 2021 · Parameterized Knowledge Transfer for Personalized Federated Learning (NeurIPS 2021 · 25 citations)
- 2022 · Layer-Wised Model Aggregation for Personalized Federated Learning (CVPR 2022 · 237 citations)
- 2023 · Tackling Data Heterogeneity in Federated Learning with Class Prototypes (AAAI 2023 · 121 citations)
- 2024 · FedAS: Bridging Inconsistency in Personalized Federated Learning (CVPR 2024 · 99 citations)
- 2025 · PFedCS: A Personalized Federated Learning Method for Enhancing Collaboration among Similar Classifiers (AAAI 2025 · 13 citations)
- 2026 · Personalized Federated Graph-Level Clustering Network (AAAI 2026 · 2 citations)
Earliest papers here
- Parameterized Knowledge Transfer for Personalized Federated Learning (NeurIPS 2021 · 25 citations)
- Personalized Federated Learning With Gaussian Processes (NeurIPS 2021 · 11 citations)
- Layer-Wised Model Aggregation for Personalized Federated Learning (CVPR 2022 · 237 citations)
What has been tried
35 papers here report 366 measured changes (a component added, removed or swapped, compared with a baseline): helped 63%, no clear effect 7%, hurt 28%, 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 Parameter Alignment and Client Synchronization | CIFAR10 (P=0.2, 0.4, 0.6, 1) · accuracy | ↑0.58; ↑1.10; ↑0.91; ↑1.09 | Helped | FedAS: Bridging Inconsistency in Personalized Federated Learning (CVPR 2024 · 99 citations) |
| Personalize parameters with the least variation rather than selecting high-change parameters | CIFAR-10 · accuracy | 79.87 vs 82.25 | Hurt | FedSelect: Personalized Federated Learning with Customized Selection of Parameters for Fine-Tuning (CVPR 2024 · 38 citations) |
| Use local-global interpolation rather than local-only feature-distribution estimates | CIFAR100 Dir(0.1) · test accuracy | .458(.08) -> .523(.08) | Helped | Personalized Federated Learning via Feature Distribution Adaptation (NeurIPS 2024 · 31 citations) |
| add server-side knowledge distillation to the global adapter | CIFAR-10 Global-test · accuracy | 53.81± 6.27 -> 59.05± 5.24 | Helped | PerAda: Parameter-Efficient Federated Learning Personalization with Generalization Guarantees (CVPR 2024 · 20 citations) |
| Use adaptive collaborator selection based on classifier distances instead of selecting none | CIFAR-100, pathological Non-IID · top-1 accuracy | Helped | PFedCS: A Personalized Federated Learning Method for Enhancing Collaboration among Similar Classifiers (AAAI 2025 · 13 citations) | |
| Remove the round-time term from the reward function | CIFAR10 Dα = 0.3 · accuracy | 83.33% ± 4.97 | Helped | PerFedRLNAS: One-for-All Personalized Federated Neural Architecture Search (AAAI 2024 · 11 citations) |
| Add the inter-client contrastive loss | Net · prediction error | Helped | Personalized Federated Learning for Spatio-Temporal Forecasting: A Dual Semantic Alignment-Based Contrastive Approach (AAAI 2025 · 8 citations) | |
| replace personalized client architectures with one uniform client model | PubMed, N=3, β=0.2, Global · accuracy | 0.73 -> 0.86 | Hurt | Large Language Models Enhanced Personalized Graph Neural Architecture Search in Federated Learning (AAAI 2025 · 7 citations) |
| Increase the larger perturbation amplitude ρlarger, then reduce it when perturbation becomes excessive | CIFAR100 Dir(0.3) · test accuracy | Mixed | Flexible Sharpness-Aware Personalized Federated Learning (AAAI 2025 · 4 citations) | |
| remove Messenger Head | Breast cancer classification · ACC | 0.9006 -> 0.8657 | Hurt | MH-pFLID: Model Heterogeneous personalized Federated Learning via Injection and Distillation for Medical Data Analysis (ICML 2024 · 2 citations) |
| add global aggregation | CIFAR-100 · accuracy | 94.41 -> 95.30 | Helped | Look Back for More: Harnessing Historical Sequential Updates for Personalized Federated Adapter Tuning (AAAI 2025 · 1 citation) |
| decrease personalization from high to low | MNIST Logit · test accuracy | 0.8123 -> 0.8312 | Helped | Understanding the Statistical Accuracy-Communication Trade-off in Personalized Federated Learning with Minimax Guarantees (ICML 2025 · 1 citation) |
What reviewers will expect
Datasets and baselines named in the 11 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 · 8 of 11 papers
- CIFAR-100 · 7 of 11 papers
- Tiny-ImageNet · 4 of 11 papers
- DomainNet · 2 of 11 papers
- MNIST · 2 of 11 papers
Baselines
- FedAvg · 6 of 11 papers
- Per-FedAvg · 4 of 11 papers
- pFedMe · 4 of 11 papers
- FedRep · 4 of 11 papers
- FedProx · 4 of 11 papers
- FedPer · 3 of 11 papers
- FedALA · 2 of 11 papers
- FedBABU · 2 of 11 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 42 papers; from OpenAlex and the papers' affiliations).
- Northeastern University · 3 papers
- Shanghai Jiao Tong University · 3 papers
- Wuhan University · 2 papers
- Nanyang Technological University · 2 papers
- The Hong Kong Polytechnic University · 2 papers
- Zhejiang University · 2 papers
- Hong Kong University of Science and Technology · 2 papers
- University of British Columbia · 2 papers
- University of California, Los Angeles · 2 papers
- Pennsylvania State University · 2 papers
Use it
Download BibTeX of 31 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
- Sparse and collaboration-graph personalization (36 papers)
- Bayesian personalized FL (35 papers)
Elsewhere in the topic
- Non-IID federated learning (198 papers)
- Federated learning with pretrained models (164 papers)
- Federated prompt and LoRA tuning (91 papers)
- Federated semi-supervised and graph learning (84 papers)
- Federated medical and multimodal learning (44 papers)
- Vertical federated learning (38 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).