Research map / Federated learning and privacy / Personalized and heterogeneous FL
FedEBA+: Towards Fair and Effective Federated Learning via Entropy-Based Model
Zhichao Wang, Lin Wang, Ye Shi, Sai Praneeth Karimireddy, Xiaoying Tang
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 FedEBA+ entropy-based aggregation with step-wise alignment instead of FedAvg | FedAvg · CIFAR-100 (Dirichlet 0.1, ResNet-18) | cross-client accuracy variance: 297.22 -> 189.06 | Improved |
| use FedEBA+ instead of FedAvg | FedAvg · CIFAR-10 (2 shards, CNN) | cross-client accuracy variance: 103.83 -> 68.71 | Improved |
| use FedEBA+ instead of FedAvg | FedAvg · CIFAR-10 (2 shards, CNN) | global accuracy: 67.79 -> 72.75 | Improved |
| use FedEBA+ instead of FedAvg | FedAvg · FashionMNIST (2 shards, MLP) | cross-client accuracy variance: 62.44 -> 43.41 | Improved |
| use FedEBA+ instead of FedAvg | FedAvg · Tiny-ImageNet (Dirichlet 0.1, MobileNet-v2) | cross-client accuracy variance: 19.62 -> 13.89 | Improved |
| use full FedEBA+ instead of practical Prac-FedEBA+ (extra communication for fair gradient) | Prac-FedEBA+ · CIFAR-10 (2 shards, CNN) | global accuracy: 69.83 -> 72.75 | Improved |
| use full FedEBA+ instead of practical Prac-FedEBA+ | Prac-FedEBA+ · CIFAR-100 (Dirichlet 0.1, ResNet-18) | cross-client accuracy variance: 231.95 -> 189.06 | Improved |
| optimize worst client with AFL instead of FedEBA+ | AFL · CIFAR-100 (Dirichlet 0.1, ResNet-18) | global accuracy: 31.98 -> 20.84 | Hurt |
| combine with Local Self-Regularization (LSR) under 50% label noise | FedEBA+ · CIFAR-10 (noisy labels 0.5) | global accuracy: 35.04 -> 61.21 | Improved |
| combine with Local Self-Regularization (LSR) under 50% label noise | FedEBA+ · FashionMNIST (noisy labels 0.5) | global accuracy: 82.03 -> 85.30 | Improved |
| increase alignment strength alpha (ablation) | FedEBA+ · hyperparameter sweep | variance and accuracy | Mixed |
| decrease temperature tau (ablation) | FedEBA+ · hyperparameter sweep | variance | Improved |
Closest papers by meaning
- Fair Federated Learning under Domain Skew with Local Consistency and Domain Diversity (CVPR 2024 · similarity 0.91)
- FairTrade: Achieving Pareto-Optimal Trade-Offs between Balanced Accuracy and Fairness in Federated Learning (AAAI 2024 · similarity 0.91)
- FedFACT: A Provable Framework for Controllable Group-Fairness Calibration in Federated Learning (NeurIPS 2025 · similarity 0.90)
- FedLF: Layer-Wise Fair Federated Learning (AAAI 2024 · similarity 0.90)
- FedAA: A Reinforcement Learning Perspective on Adaptive Aggregation for Fair and Robust Federated Learning (AAAI 2025 · similarity 0.90)
- Demystifying Local & Global Fairness Trade-offs in Federated Learning Using Partial Information Decomposition (ICLR 2024 · similarity 0.90)
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.
