Research map / Federated learning and privacy / Communication-efficient FL
A Random Projection Approach to Personalized Federated Learning: Enhancing Communication Efficiency, Robustness, and Fairness
Yuze Han, Xiang Li, Shiyun Lin, Zhihua Zhang
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 |
|---|---|---|---|
| use random subspace projection with L1 regularization (lp-proj-1) instead of FedAvg | personalized FL · ImageNet ResNet34 | test accuracy: 0.0458 (0.0004) -> 0.8403 (0.0071) | Improved |
| use random projection (lp-proj-1) instead of pFedMe | personalized FL · ImageNet ResNet34 | communication bytes: 2.582 × 1011 -> 1.2 × 107 | Improved |
| use lp-proj-1 instead of pFedMe | personalized FL · ImageNet ResNet34 | test accuracy: 0.5009 (0.0130) -> 0.8403 (0.0071) | Improved |
| use lp-proj-1 instead of LBGM | personalized compressed FL · Synthetic(0, 0) target accuracy 0.6 | bytes used: 12200 -> 4620 | Improved |
| use lp-proj-1 instead of FedAvg under byte budget | personalized FL · Synthetic(0, 0), 328020 bytes budget | test accuracy: 0.625 -> 0.885 | Improved |
| use lp-proj-1 instead of best competing method | personalized FL · EMNIST target accuracy 0.7 | communication bytes: 79x and 1320x | Improved |
| use lp-proj instead of other methods under same-value and Gaussian attacks | robust FL · EMNIST/Synthetic Byzantine attacks | benign client test accuracy | Improved |
| use lp-proj-1 under sign-flipping attack with more than 20% malicious clients | robust FL · sign-flipping attack | benign client test accuracy | Improved |
| use lp-proj-1 instead of RSA | fair FL · CIFAR | test accuracy (%): 77.68 -> 79.22 | Improved |
| use lp-proj-2 instead of Ditto | fair FL · EMNIST | test accuracy (%): 90.89 -> 90.70 | No clear effect |
| use lp-proj instead of Ditto | personalized FL · EMNIST test accuracy curves | test accuracy | No clear effect |
| add block-diagonal projection with 50 blocks | lp-proj · ImageNet ResNet34 | space and computation complexity | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- Ditto — Ditto: Fair and Robust Federated Learning Through Personalization (ICML 2021) · better on 0 of 1 shared result
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
- Understanding the Statistical Accuracy-Communication Trade-off in Personalized Federated Learning with Minimax Guarantees (ICML 2025 · similarity 0.91)
- Fine-Tuning Personalization in Federated Learning to Mitigate Adversarial Clients (NeurIPS 2024 · similarity 0.89)
- Flexible Sharpness-Aware Personalized Federated Learning (AAAI 2025 · similarity 0.88)
- Decentralized Directed Collaboration for Personalized Federated Learning (CVPR 2024 · similarity 0.88)
- Adaptive Latent-Space Constraints in Personalized Federated Learning (NeurIPS 2025 · similarity 0.88)
- PeFLL: Personalized Federated Learning by Learning to Learn (ICLR 2024 · similarity 0.88)
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.
