Research map / Federated learning and privacy / Personalized and heterogeneous FL
FedDisco: Federated Learning with Discrepancy-Aware Collaboration
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 |
|---|---|---|---|
| Replace dataset-size-only aggregation with discrepancy-aware weights (ablation) | FedAvg on partial-participation CIFAR-10 NIID-2 · CIFAR-10 NIID-2, partial participation | accuracy (%): 54.27 -> 61.59; +7.32 | Helped |
| Replace dataset-size-only aggregation with discrepancy-aware weights (ablation) | FedAvg on partial-participation CIFAR-10 NIID-2 · CIFAR-10 NIID-2, partial participation | rounds to 55% accuracy: 58 -> 20; ↓38, ↓65.52% | Helped |
| Add Disco reweighting (ablation) | FedAvg · AG News full participation | accuracy: 82.03 -> 84.09 | Helped |
| Add Disco reweighting (ablation) | FedProx · AG News full participation | accuracy: 79.89 -> 80.81 | Helped |
| Add Disco reweighting (ablation) | FedAvg · AG News partial participation | accuracy: 50.88 -> 57.71 | Helped |
| Add Disco reweighting (ablation) | FedProx · AG News partial participation | accuracy: 58.22 -> 62.86 | Helped |
| Use KL-Divergence rather than L1 discrepancy (ablation) | FedDisco · CIFAR-10 NIID-1 | accuracy: 69.43 -> 70.05 | No clear effect |
| Use L2 rather than L1 discrepancy (ablation) | FedDisco · CIFAR-10 NIID-1 | accuracy: 69.43 -> 69.99 | No clear effect |
| Use cosine similarity rather than L1 discrepancy (ablation) | FedDisco · CIFAR-10 NIID-1 | accuracy: 69.43 -> 69.90 | No clear effect |
| Add discrepancy-aware aggregation (ablation) | FedAvg · CIFAR-100 NIID-1 | accuracy (%): 57.28 -> 58.28 | Helped |
| Add discrepancy-aware aggregation (ablation) | FedDyn · CIFAR-100 NIID-1 | accuracy (%): 62.29 -> 62.83 | Helped |
| Add discrepancy-aware aggregation (ablation) | FedAvg · CIFAR-10, device heterogeneity with stragglers | accuracy: 60.21% -> 63.20% | Helped |
Closest papers by meaning
- FedAWA: Adaptive Optimization of Aggregation Weights in Federated Learning Using Client Vectors (CVPR 2025 · similarity 0.89)
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning (NeurIPS 2025 · similarity 0.89)
- FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning (NeurIPS 2025 · similarity 0.89)
- FedGMKD: An Efficient Prototype Federated Learning Framework through Knowledge Distillation and Discrepancy-Aware Aggregation (NeurIPS 2024 · similarity 0.88)
- Progressive Distribution Matching for Federated Semi-Supervised Learning (AAAI 2025 · similarity 0.88)
- FLHetBench: Benchmarking Device and State Heterogeneity in Federated Learning (CVPR 2024 · similarity 0.88)
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.