Research map / Federated learning and privacy / Personalized and heterogeneous FL
A Bayesian Framework for Clustered Federated Learning
Peng Wu, Tales Imbiriba, Pau Closas
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 |
|---|---|---|---|
| explore multiple association hypotheses (MH-3) instead of greedy association (G) (ablation) | BCFL · CIFAR-10, Label (4,10,0.1) | accuracy (%): 64.73 -> 74.35 | Improved |
| explore multiple association hypotheses (MH-3) instead of greedy association (G) (ablation) | BCFL · Digits-Five, Feature (5,2) | accuracy (%): 94.35 -> 95.39 | Improved |
| increase maximum hypotheses from 3 to 6 in BCFL-MH (ablation) | BCFL-MH · AmazonReview, Feature (4,2) | accuracy (%): 88.74 -> 88.16 | Hurt |
| increase maximum hypotheses from 3 to 6 in BCFL-MH (ablation) | BCFL-MH · Digits-Five, Feature (5,2) | accuracy (%): 95.39 -> 96.02 | Improved |
| increase candidate associations from 3 to 6 in BCFL-C (ablation) | BCFL-C · CIFAR-10, Label (4,10,0.1) | accuracy (%): 72.12 -> 68.73 | Hurt |
| use BCFL-MH-3 instead of FedAvg | FedAvg · CIFAR-10, Label (4,10,0.1) | accuracy (%): 39.89 -> 74.35 | Improved |
| use FedAvg single model instead of BCFL-G | BCFL-G · AmazonReview, Feature (4,2) | accuracy (%): 87.71 -> 87.53 | No clear effect |
| add warm-up stage to BCFL-G (ablation) | BCFL-G · CIFAR-10, Label (4,10,0.1) | accuracy (%): 64.73 -> 72.95 | Improved |
| add warm-up stage to BCFL-G (ablation) | BCFL-G · Fashion-MNIST, Label (0,40,0.1) | accuracy (%): 92.79 -> 92.46 | Hurt |
| add warm-up stage to BCFL-C-3 (ablation) | BCFL-C-3 · Fashion-MNIST, Label (0,40,0.1) | accuracy (%): 91.24 -> 92.39 | Improved |
| add warm-up stage to BCFL-MH-6 (ablation) | BCFL-MH-6 · Fashion-MNIST, Label (0,40,0.1) | accuracy (%): 94.29 -> 94.22 | No clear effect |
| assume fewer clusters than true (K=2 instead of 4) (ablation) | BCFL-MH-3 · Fashion-MNIST, Label (4,10,0.1) | accuracy (%): 97.13 -> 94.39 | Hurt |
Closest papers by meaning
- Enhancing Clustered Federated Learning: Integration of Strategies and Improved Methodologies (ICLR 2025 · similarity 0.92)
- Clustered Federated Learning via Gradient-based Partitioning (ICML 2024 · similarity 0.90)
- A Bayesian Approach for Personalized Federated Learning in Heterogeneous Settings (NeurIPS 2024 · similarity 0.90)
- Interaction-Aware Gaussian Weighting for Clustered Federated Learning (ICML 2025 · similarity 0.89)
- FLUX: Efficient Descriptor-Driven Clustered Federated Learning under Arbitrary Distribution Shifts (NeurIPS 2025 · similarity 0.88)
- Harnessing Heterogeneous Statistical Strength for Personalized Federated Learning via Hierarchical Bayesian Inference (ICML 2025 · 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.
