Research map / Federated learning and privacy / Communication-efficient FL
FedMix: Approximation of Mixup under Mean Augmented Federated Learning
Tehrim Yoon, Sumin Shin, Sung Ju Hwang, Eunho Yang
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 derivative-based FedMix with averaged external data instead of FedAvg | Federated learning on CIFAR10 · CIFAR10 | test accuracy: 73.8 -> 81.2 | Improved |
| Use derivative-based FedMix with averaged external data instead of FedAvg | Federated learning on CIFAR100 · CIFAR100 | test accuracy: 50.4 -> 56.7 | Improved |
| Use derivative-based FedMix with averaged external data instead of FedAvg | Federated learning on FEMNIST · FEMNIST | test accuracy: 85.3 -> 86.5 | Improved |
| Add FedMix to FedProx | FedProx · CIFAR10 | test accuracy: 77.3 -> 78.9 | Improved |
| Use NaiveMix instead of FedAvg | Federated learning · CIFAR10 | test accuracy: 73.8 -> 77.4 | Improved |
| Use FedMix instead of NaiveMix | MAFL · CIFAR10 | test accuracy: 77.4 -> 81.2 | Improved |
| Use FedMix instead of NaiveMix | MAFL · Shakespeare | test accuracy: 56.9 -> 56.9 | No clear effect |
| Approximate global Mixup with FedMix rather than perform Global Mixup | CIFAR10 federated learning · CIFAR10 | test accuracy: 88.2 -> 81.2 | Hurt |
| Use FedMix when clients have data from only 2 classes instead of 10 | Federated learning · CIFAR10 | test accuracy: 89.1; 81.2 | Mixed |
| Use FedMix with partial participation (K/N=0.1) instead of full participation (K/N=1.0) | Federated learning · CIFAR10 | test accuracy: 84.3 -> 74.7 | Hurt |
| Use larger mixup ratio λ=0.5 rather than λ=0.05 (ablation) | FedMix · CIFAR10 | test accuracy: 81.2 -> 67.1 | Hurt |
| Use same-class split rather than random split for averaging (ablation) | FedMix · CIFAR10 | test accuracy: 81.2 -> 78.8 | Hurt |
Later papers that compared with this method
3 later accepted papers report FedMix as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- FedCFA: Alleviating Simpson’s Paradox in Model Aggregation with Counterfactual Federated Learning (AAAI 2025 · better on 8 of 8 shared results)
- FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise (AAAI 2024 · better on 1 of 1 shared result)
- FedFA: Federated Feature Augmentation (ICLR 2023 · better on 7 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
- FedAvP: Augment Local Data via Shared Policy in Federated Learning (NeurIPS 2024 · similarity 0.87)
- pFedMxF: Personalized Federated Class-Incremental Learning with Mixture of Frequency Aggregation (CVPR 2025 · similarity 0.87)
- Low Precision Local Training is Enough for Federated Learning (NeurIPS 2024 · similarity 0.87)
- A Simple Data Augmentation for Feature Distribution Skewed Federated Learning (CVPR 2025 · similarity 0.86)
- One Arrow, Two Hawks: Sharpness-aware Minimization for Federated Learning via Global Model Trajectory (ICML 2025 · similarity 0.86)
- FedSOL: Stabilized Orthogonal Learning with Proximal Restrictions in Federated Learning (CVPR 2024 · similarity 0.86)
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.
