Research map / Federated learning and privacy / Personalized and heterogeneous FL
CalFAT: Calibrated Federated Adversarial Training with Label Skewness
Chen Chen, Yuchen Liu, Xingjun Ma, Lingjuan Lyu
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 |
|---|---|---|---|
| replace standard CE with calibrated logits loss in all FAT methods (MixFAT) (ablation) | MixFAT · CIFAR10 | natural accuracy: 53.35 -> 60.23 | Improved |
| add calibration loss to FedMART (ablation) | FedMART · CIFAR10 | PGD-20 robust accuracy: 18.10 -> 27.19 | Improved |
| calibration loss instead of LogitAdj (ablation) | MixFAT · CIFAR10 | natural accuracy: 57.53 -> 60.23 | Improved |
| calibration loss instead of RoBal (ablation) | MixFAT · CIFAR10 | PGD-20 robust accuracy: 27.86 -> 28.67 | Improved |
| add calibrated CE loss (ablation) | CalFAT · CIFAR10 beta=0.05 | PGD-20 robust accuracy: 20.55 -> 29.49 | Improved |
| add calibrated KL loss (ablation) | CalFAT · CIFAR10 beta=0.05 | PGD-20 robust accuracy: 27.59 -> 29.49 | Improved |
| increase adversarial data ratio from 0 to 1 (ablation) | CalFAT · CIFAR10 | PGD-20 robust accuracy: 3.47 -> 31.19 | Improved |
| CalFAT vs best baseline FedGAIRAT | CalFAT · SVHN | natural accuracy: 58.41 -> 84.15 | Improved |
| CalFAT vs best baseline | CalFAT · ImageNet subset | PGD-20 robust accuracy: 19.18 -> 19.97 | Improved |
Later papers that compared with this method
1 later accepted paper reports CalFAT as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- FedCART: Tackling Long-Tailed Distributions in Federated Adversarial Training via Classifier Refinement (CVPR 2026 · better on 5 of 7 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
- Exploiting Label Skews in Federated Learning with Model Concatenation (AAAI 2024 · similarity 0.86)
- Exploring Vacant Classes in Label-Skewed Federated Learning (AAAI 2025 · similarity 0.86)
- Recovering Labels from Local Updates in Federated Learning (ICML 2024 · similarity 0.84)
- Byzantine-Robust Federated Learning: Impact of Client Subsampling and Local Updates (ICML 2024 · similarity 0.83)
- How Does the Smoothness Approximation Method Facilitate Generalization for Federated Adversarial Learning? (AAAI 2025 · similarity 0.83)
- Federated Learning with Extremely Noisy Clients via Negative Distillation (AAAI 2024 · similarity 0.83)
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.
