Research map / Federated learning and privacy / Personalized and heterogeneous FL
Federated Active Learning Under Extreme Non-IID and Global Class Imbalance
Chen-Chen Zong, Sheng-Jun Huang
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 adaptively selected query model instead of the complementary one (ablation) | FairFAL · ε=0.1, imbalance 20 | final test accuracy (%): 58.49 -> 59.33 | Improved |
| use adaptively selected query model instead of the complementary one (ablation) | FairFAL · ε=100, imbalance 20 | final test accuracy (%): 61.89 -> 63.65 | Improved |
| build prototypes from global features instead of local features (ablation) | FairFAL · ε=0.1, imbalance 20 | final test accuracy (%): 59.14 -> 59.95 | Improved |
| build prototypes from global features instead of local features (ablation) | FairFAL · ε=100, imbalance 20 | final test accuracy (%): 63.39 -> 64.02 | Improved |
| add two-stage uncertainty-diversity sampling (alpha=4) after class-wise sampling (ablation) | FairFAL · ε=0.1, imbalance 20 | final test accuracy (%): 59.95 -> 60.44 | Improved |
| vary candidate pool ratio alpha among 2, 3, 4 (ablation) | FairFAL · ε=0.1 and ε=100, imbalance 20 | final test accuracy (%): 60.61 / 60.38 / 60.44 | No clear effect |
| query with global model vs local model under Entropy, homogeneous clients with severe imbalance (ε=100, imbalance 20) | Entropy sampling · CIFAR-10 | AULC: +106 pp | Improved |
| query with global model vs local model under Entropy, heterogeneous clients (ε=0.1, imbalance 1) | Entropy sampling · CIFAR-10 | AULC: local wins by 201 pp | Hurt |
| use global model rather than local for Coreset sampling | Coreset sampling · CIFAR-10 | AULC: +21 to +92 pp | Improved |
| FairFAL vs best baseline on OctMNIST | FairFAL · OctMNIST (ε=0.1) | test accuracy (%): 70.40 -> 72.80 | Improved |
Closest papers by meaning
- Efficient Heterogeneity-Aware Federated Active Data Selection (ICML 2025 · similarity 0.89)
- FedFACT: A Provable Framework for Controllable Group-Fairness Calibration in Federated Learning (NeurIPS 2025 · similarity 0.87)
- FairTrade: Achieving Pareto-Optimal Trade-Offs between Balanced Accuracy and Fairness in Federated Learning (AAAI 2024 · similarity 0.86)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.86)
- FedLF: Layer-Wise Fair Federated Learning (AAAI 2024 · similarity 0.86)
- LoGoFair: Post-Processing for Local and Global Fairness in Federated Learning (AAAI 2025 · similarity 0.85)
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.
