Research map / Federated learning and privacy / Personalized and heterogeneous FL
Sparse and collaboration-graph personalization
Adapt models to clients via sparse adaptation, latent constraints and inferred collaboration graphs.
36 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.
14 papers in 2023–24 and 14 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 5.6% to 3.6%.
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 · Federated Multi-Task Learning under a Mixture of Distributions (NeurIPS 2021 · 91 citations)
- 2022 · FedSoft: Soft Clustered Federated Learning with Proximal Local Updating (AAAI 2022 · 116 citations)
- 2023 · Personalized Subgraph Federated Learning (ICML 2023 · 11 citations)
- 2024 · Dual-Personalizing Adapter for Federated Foundation Models (NeurIPS 2024 · 13 citations)
- 2025 · Patient-Level Anatomy Meets Scanning-Level Physics: Personalized Federated Low-Dose CT Denoising Empowered by Large Language Model (CVPR 2025 · 18 citations)
- 2026 · Federated Context-Aware Personalized Recommendation (AAAI 2026 · 1 citation)
Earliest papers here
- Federated Multi-Task Learning under a Mixture of Distributions (NeurIPS 2021 · 91 citations)
- Exploiting Shared Representations for Personalized Federated Learning (ICML 2021 · 69 citations)
- PartialFed: Cross-Domain Personalized Federated Learning via Partial Initialization (NeurIPS 2021 · 48 citations)
What has been tried
30 papers here report 292 measured changes (a component added, removed or swapped, compared with a baseline): helped 69%, no clear effect 5%, hurt 22%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| add scanning-level prompt to the generic model | LDCT clients #1–#8, average · PSNR | 34.03 -> 36.82 | Helped | Patient-Level Anatomy Meets Scanning-Level Physics: Personalized Federated Low-Dose CT Denoising Empowered by Large Language Model (CVPR 2025 · 18 citations) |
| Add personalized gating rather than fixed encoder weight w=0.25 | ML-100K · HR@20 | 0.1213 (5.75%↓) | Helped | Personalized Federated Collaborative Filtering: A Variational AutoEncoder Approach (AAAI 2025 · 15 citations) |
| add instance-wise dynamic weighting | Fed Dataset 1 and Fed Dataset 2 personalization · ROGUE-1 | 79.06 -> 79.48; 73.17 -> 73.48 | Helped | Dual-Personalizing Adapter for Federated Foundation Models (NeurIPS 2024 · 13 citations) |
| Add group prompts to a shared-prompt model | CIFAR-100, s=10 · worst local accuracy | 63.38 -> 73.85 | Helped | Unlocking the Potential of Prompt-Tuning in Bridging Generalized and Personalized Federated Learning (CVPR 2024 · 12 citations) |
| Add selected saved server models (M=8 instead of M=0, b=10) | CIFAR-10 · accuracy | 65.55% ± 8.77% -> 66.54% ± 8.08% | Helped | Personalized Federated Learning with Mixture of Models for Adaptive Prediction and Model Fine-Tuning (NeurIPS 2024 · 11 citations) |
| Use additive personalized item embeddings rather than a shared-only item embedding (FedRAP-C) | ML-100K · HR@10 | Helped | Federated Recommendation with Additive Personalization (ICLR 2024 · 8 citations) | |
| Replace CLIA with the meta-classifier attack | CIFAR-10, α=0.5, client 0 · ASR | 37.49 -> 74.28 | Helped | Personalized Label Inference Attack in Federated Transfer Learning via Contrastive Meta Learning (AAAI 2025 · 4 citations) |
| add AAC instead of using a unified classifier | CIFAR-100, QBLI α=6 · Avg. accuracy (%) | 74.1 -> 80.7 | Helped | pFedMxF: Personalized Federated Class-Incremental Learning with Mixture of Frequency Aggregation (CVPR 2025 · 4 citations) |
| replace dense ETF with SSE-C | CIFAR-100-LT (α=0.5, IF=100) · GM accuracy | 0.3825 -> 0.4175 | Helped | FedLoGe: Joint Local and Generic Federated Learning under Long-tailed Data (ICLR 2024 · 3 citations) |
| Add client-wise feature disentanglement | Mini-ImageNet, FSL Homogeneous, 5-way 1-shot · accuracy | 51.32±1.41 -> 51.96±1.75 | Helped | FedFSL-CFRD: Personalized Federated Few-Shot Learning with Collaborative Feature Representation Disentanglement (AAAI 2025 · 2 citations) |
| add dynamic cooperation (aggregation over coalitions) vs. local-only training | EMNIST-LTP · average accuracy | 12.3 to 52.5 | Helped | Decentralized Dynamic Cooperation of Personalized Models for Federated Continual Learning (NeurIPS 2025 · 1 citation) |
| increase the number of constrained latent layers beyond depth 1 | CIFAR-10 α=0.1 · accuracy | Hurt | Adaptive Latent-Space Constraints in Personalized Federated Learning (NeurIPS 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 9 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 · 3 of 9 papers
- CIFAR-100 · 3 of 9 papers
- Fashion-MNIST · 2 of 9 papers
- EMNIST · 2 of 9 papers
Baselines
- FedAvg · 6 of 9 papers
- FedProx · 3 of 9 papers
- FedProto · 2 of 9 papers
- PerFedAvg · 2 of 9 papers
- FedPer · 2 of 9 papers
- MOON · 2 of 9 papers
- GLFC · 2 of 9 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 28 papers; from OpenAlex and the papers' affiliations).
- University of Technology Sydney · 3 papers
- University of Maryland, College Park · 2 papers
- Tsinghua University · 2 papers
- Microsoft · 2 papers
- University of Cambridge · 2 papers
Use it
Download BibTeX of 32 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
- Personalization with global-local trade-off (48 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).