Research map / Federated learning and privacy / Personalized and heterogeneous FL
Robust Federated Learning: The Case of Affine Distribution Shifts
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 FedRobust instead of FedAvg under affine distribution shifts | CIFAR-10 models (AlexNet, Inception-Net, ResNet) · CIFAR-10, affine distribution shifts | test accuracy: up to 54% | Helped |
| Use FedRobust instead of distributed FGM or PGD training under affine distribution shifts | CIFAR-10 models (AlexNet, Inception-Net, ResNet) · CIFAR-10, affine distribution shifts | test accuracy | Helped |
| Use FedRobust instead of distributed PGD training | CIFAR-10 models · CIFAR-10 training | training speed: 4 \times | Helped |
| Use FedRobust instead of FGM or PGD adversarial training | CIFAR-10 models · CIFAR-10, PGD perturbations | test accuracy | No clear effect |
| Use FedRobust instead of FedAvg and adversarial-training baselines | MNIST models · MNIST, affine distribution shifts | test accuracy | Helped |
| Use FedRobust instead of FGM or PGD adversarial training | MNIST models · MNIST, PGD perturbations | test accuracy | No clear effect |
| Apply an affine distribution shift to test samples | FedAvg classifier · Image recognition tasks, unseen test user | classifier performance | Hurt |
Closest papers by meaning
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning (NeurIPS 2025 · similarity 0.88)
- Byzantine-Robust Federated Learning: Impact of Client Subsampling and Local Updates (ICML 2024 · similarity 0.88)
- FedWon: Triumphing Multi-domain Federated Learning Without Normalization (ICLR 2024 · similarity 0.88)
- FLUX: Efficient Descriptor-Driven Clustered Federated Learning under Arbitrary Distribution Shifts (NeurIPS 2025 · similarity 0.87)
- Federated Learning with Sample-level Client Drift Mitigation (AAAI 2025 · similarity 0.87)
- Certifiably Robust Model Evaluation in Federated Learning under Meta-Distributional Shifts (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.