Research map / Federated learning and privacy / Personalized and heterogeneous FL
Robust Asymmetric Heterogeneous Federated Learning With Corrupted Clients
Xiuwen Fang, Mang Ye, Bo Du
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 AsymHFL instead of no collaboration (ablation) | local training · CIFAR-10-C heterogeneous FL (xi=0.5, test clean) | average test accuracy (%): 62.94 -> 79.13 | Improved |
| add AsymHFL instead of no collaboration (ablation) | local training · CIFAR-10-C heterogeneous FL (xi=0.5, test corrupted) | average test accuracy (%): 58.01 -> 74.44 | Improved |
| add symmetric HFL knowledge sharing (ablation) | local training · CIFAR-10-C heterogeneous FL (xi=0.5, test clean) | average test accuracy (%): 62.94 -> 60.13 | Hurt |
| add symmetric HFL knowledge sharing (ablation) | local training · CIFAR-10-C heterogeneous FL (xi=1, test corrupted) | average test accuracy (%): 58.10 -> 57.71 | No clear effect |
| add AsymHFL instead of symmetric HFL (ablation) | HFL · CIFAR-10-C heterogeneous FL (xi=1, test corrupted) | average test accuracy (%): 57.71 -> 73.96 | Improved |
| add random mixed augmentation with JSD consistency (Aug) (ablation) | AsymHFL · CIFAR-10-C heterogeneous FL (xi=0, test corrupted) | average test accuracy (%): 65.93 -> 75.84 | Improved |
| add random mixed augmentation with JSD consistency (Aug) (ablation) | AsymHFL · CIFAR-10-C heterogeneous FL (xi=0.5, test clean) | average test accuracy (%): 79.13 -> 80.14 | Improved |
| add DCL on top of Aug (ablation) | AsymHFL+Aug · CIFAR-10-C heterogeneous FL (xi=0, test clean) | average test accuracy (%): 82.92 -> 84.40 | Improved |
| add DCL on top of Aug (ablation) | AsymHFL+Aug · CIFAR-10-C heterogeneous FL (xi=0, test corrupted) | average test accuracy (%): 75.84 -> 77.43 | Improved |
| add DCL on top of Aug (ablation) | AsymHFL+Aug · CIFAR-10-C heterogeneous FL (xi=1, test corrupted) | average test accuracy (%): 76.08 -> 77.59 | Improved |
| use standard SupCon with complex augmentations instead of DCL (ablation) | AsymHFL+Aug · CIFAR-10-C heterogeneous FL (xi=0, test corrupted) | average test accuracy (%): 75.84 -> 74.37 | Hurt |
| use standard SupCon with complex augmentations instead of DCL (ablation) | AsymHFL+Aug · CIFAR-10-C heterogeneous FL (xi=0, test clean) | average test accuracy (%): 82.92 -> 83.31 | No clear effect |
Closest papers by meaning
- Relaxed Contrastive Learning for Federated Learning (CVPR 2024 · similarity 0.89)
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning (NeurIPS 2025 · similarity 0.88)
- Federated Learning with Sample-level Client Drift Mitigation (AAAI 2025 · similarity 0.88)
- Byzantine-Robust Federated Learning: Impact of Client Subsampling and Local Updates (ICML 2024 · similarity 0.87)
- FedAdaVR: Adaptive Variance Reduction for Robust Federated Learning under Limited Client Participation (NeurIPS 2026 · similarity 0.87)
- Scalable and Trustworthy Learning in Heterogeneous Networks (AAAI 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.
