Research map / Federated learning and privacy / Personalized and heterogeneous FL
Cooperative Pseudo Labeling for Unsupervised Federated Classification
Kuangpu Guo, Lijun Sheng, Yongcan Yu, Jian Liang, Zilei Wang, Ran He
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 global allocation of pseudo labels (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 59.76 -> 69.60 | Improved |
| add entropy filter on top of global allocation (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 69.60 -> 72.28 | Improved |
| add confidence filter on top of global allocation (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 69.60 -> 73.66 | Improved |
| add confidence filter to entropy filter plus global allocation (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 72.28 -> 75.76 | Improved |
| aggregate only visual prompts instead of both modalities (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 61.68 -> 75.76 | Improved |
| aggregate only visual prompts instead of only textual prompts (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 67.12 -> 75.76 | Improved |
| aggregate visual prompts instead of keeping both prompts local (ablation) | FedCoPL · RESISC45 (Dirichlet beta=0.1) | accuracy: 61.72 -> 75.76 | Improved |
| replace FPL pseudo labeling with CoPL | FedOPT · RESISC45 (Dirichlet beta=0.1) | accuracy: 57.76 -> 74.85 | Improved |
| replace FPL pseudo labeling with CoPL | PromptFL · DTD (Dirichlet beta=0.1) | accuracy: 45.79 -> 47.34 | Improved |
| use CPL instead of FPL pseudo labeling | FedOPT · DTD (Dirichlet beta=0.1) | accuracy: 51.97 -> 37.43 | Hurt |
| use CPL instead of FPL pseudo labeling | PromptFL · DTD and CIFAR10 (Dirichlet beta=0.1) | accuracy: DTD 45.79 -> 44.84; CIFAR10 87.51 -> 88.90 | Mixed |
| raise confidence threshold tau1 from 0.50 to 0.90 (ablation) | FedCoPL · DTD (Dirichlet beta=0.1) | accuracy: 60.89 -> 36.16 | Hurt |
Closest papers by meaning
- CLIP-Guided Federated Learning on Heterogeneity and Long-Tailed Data (AAAI 2024 · similarity 0.89)
- Global and Local Prompts Cooperation via Optimal Transport for Federated Learning (CVPR 2024 · similarity 0.87)
- Language-Guided Transformer for Federated Multi-Label Classification (AAAI 2024 · similarity 0.87)
- (FL)$^2$: Overcoming Few Labels in Federated Semi-Supervised Learning (NeurIPS 2024 · similarity 0.87)
- Mixture of Experts Made Personalized: Federated Prompt Learning for Vision-Language Models (ICLR 2025 · similarity 0.86)
- Federated Learning from Vision-Language Foundation Models: Theoretical Analysis and Method (NeurIPS 2024 · similarity 0.86)
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.
