Research map / Federated learning and privacy / Personalized and heterogeneous FL
FedCD: Towards Consolidated Distillation for Heterogeneous Federated Learning
Yichen Li, Hang Su, Huifa Li, Haolin Yang, Xinlin Zhuang, Haochen Xue, Haozhao Wang, Imran Razzak
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 |
|---|---|---|---|
| add logit-based distillation loss alongside feature loss (ablation) | FedCD · CIFAR-100 (alpha=0.1) | accuracy: 58.97 -> 64.77 | Improved |
| add feature-based distillation loss alongside logit loss (ablation) | FedCD · CIFAR-10 (alpha=0.1) | accuracy: 86.46 -> 89.18 | Improved |
| add cross-layer adaptive feature fusion (ablation) | FedCD · CIFAR-10 (alpha=0.1) | accuracy: 84.99 -> 89.18 | Improved |
| add GMM distribution modeling of fused features (ablation) | FedCD · CIFAR-10 (alpha=1.0) | accuracy: 92.88 -> 95.52 | Improved |
| add GMM distribution modeling of fused features (ablation) | FedCD · CIFAR-100 (alpha=0.1) | accuracy: 58.84 -> 64.77 | Improved |
| naively sum logit and feature distillation losses (no fusion or GMM) | ensemble distillation · CIFAR-10/CIFAR-100 | accuracy | Mixed |
| use FedCD instead of best baseline (DaFKD) | federated distillation · CIFAR-100 (alpha=0.1) | accuracy: 62.01 -> 64.77 | Improved |
| use FedCD instead of FedFD under model heterogeneity | federated distillation · CIFAR-10 (alpha=1.0, model heterogeneity a-c-d-f-i) | accuracy: 86.12 -> 87.84 | Improved |
| use FedCD for faster convergence to 80% accuracy | federated distillation · CIFAR-10 (alpha=1.0, rounds to 80% accuracy) | communication rounds: 46.50 (FedFusion) vs 49.50 (FedCD) | Mixed |
| vary lambda and mixture components J (ablation) | FedCD · CIFAR-100 | accuracy | No clear effect |
| increase active client ratio r (ablation) | FedCD · CIFAR-10/CIFAR-100 | accuracy | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DaFKD — DaFKD: Domain-Aware Federated Knowledge Distillation (CVPR 2023) · better on 3 of 3 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
- Feature Distillation is the Better Choice for Model-Heterogeneous Federated Learning (NeurIPS 2025 · similarity 0.93)
- HYDRA-FL: Hybrid Knowledge Distillation for Robust and Accurate Federated Learning (NeurIPS 2024 · similarity 0.90)
- Efficient Federated Learning via Clients-to-Server Knowledge Distillation (Student Abstract) (AAAI 2025 · similarity 0.89)
- Towards Diverse Device Heterogeneous Federated Learning via Task Arithmetic Knowledge Integration (NeurIPS 2024 · similarity 0.88)
- FedGMKD: An Efficient Prototype Federated Learning Framework through Knowledge Distillation and Discrepancy-Aware Aggregation (NeurIPS 2024 · similarity 0.88)
- Resource-Aware Federated Self-Supervised Learning with Global Class Representations (NeurIPS 2024 · 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.
