Research map / Federated learning and privacy / Personalized and heterogeneous FL
Out-of-Distribution Generalization of Federated Learning via Implicit Invariant Relationships
Yaming Guo, Kai Guo, Xiaofeng Cao, Tieru Wu, Yi Chang
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 |
|---|---|---|---|
| Add inter-client gradient alignment penalty (FEDIIR) to FedAvg local objective | FEDIIR vs FEDAVG · Small number of clients, four datasets average | average test accuracy (%): +0.5 (80.6 to 81.1) | Improved |
| Add inter-client gradient alignment penalty (FEDIIR) to FedAvg local objective | FEDIIR vs FEDAVG · 50 clients with limited communication, average | average test accuracy (%): +5.7 (71.3 to 77.0) | Improved |
| Use FEDIIR instead of explicit invariant methods (FEDADG, FEDSR) | FEDIIR · 50 clients, VLCS and OfficeHome | average test accuracy (%): about 14 points on VLCS and 8 on OfficeHome vs baselines (as stated) | Improved |
| Use FEDIIR instead of baselines in small-client setting | FEDIIR · Small number of clients, VLCS | average test accuracy (%): 76.6 vs FEDADG 77.1 | Hurt |
| Use FEDIIR instead of baselines in small-client setting | FEDIIR · Small number of clients, PACS and OfficeHome | average test accuracy (%): PACS 83.7 vs FEDSR 83.4; OfficeHome 69.2 vs FEDSR 69.1 | Mixed |
| Increase gradient-alignment penalty gamma from 0 to 0.1 (ablation) | FEDIIR · 50 clients, VLCS/PACS/OfficeHome sensitivity | test accuracy (%): better than FEDAVG for gamma in {0.0001,0.001,0.01}, worse at gamma=0.1 | Mixed |
| Increase number of participating clients (3 to 100) for baselines | FEDAVG, FEDADG, FEDSR · Massively distributed clients | test accuracy (%): baseline effectiveness noticeably decreases as clients increase | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FEDSR — FedSR: A Simple and Effective Domain Generalization Method for Federated Learning (NeurIPS 2022) · better on 10 of 10 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
- Generalization in Federated Learning: A Conditional Mutual Information Framework (ICML 2025 · similarity 0.89)
- FedImpro: Measuring and Improving Client Update in Federated Learning (ICLR 2024 · similarity 0.88)
- Beyond the Federation: Topology-aware Federated Learning for Generalization to Unseen Clients (ICML 2024 · similarity 0.88)
- Causally Motivated Personalized Federated Invariant Learning with Shortcut-Averse Information-Theoretic Regularization (ICML 2024 · similarity 0.88)
- Covariances for Free: Exploiting Mean Distributions for Training-free Federated Learning (NeurIPS 2025 · similarity 0.87)
- Causality Inspired Federated Learning for OOD Generalization (ICML 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.
