Research map / Federated learning and privacy / Personalized and heterogeneous FL
ProxyFL: A Proxy-Guided Framework for Federated Semi-Supervised Learning
Duowen Chen, Yan Wang
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 Proxy Tuning to FixMatch-GPL baseline (ablation) | ProxyFL · CIFAR-100 (alpha=0.1) | accuracy: 48.96 -> 54.86 | Improved |
| add ICPL to FixMatch-GPL baseline (ablation) | ProxyFL · CIFAR-100 (alpha=0.1) | accuracy: 48.96 -> 57.21 | Improved |
| combine GPT and ICPL (ablation) | ProxyFL · CIFAR-100 (alpha=0.1) | accuracy: 57.21 -> 57.50 | Improved |
| combine GPT and ICPL (ablation) | ProxyFL · CINIC-10 (alpha=1) | accuracy: 79.31 -> 79.59 | Improved |
| use indecisive-category set instead of directly pseudo-labeling all low-confidence samples (ablation) | ICPL · SVHN (alpha=0.1) | accuracy: 83.99 -> 94.82 | Improved |
| directly include all low-confidence samples with pseudo-labels (GPL-ALL vs GPL) (ablation) | FixMatch-GPL · CIFAR-100 (alpha=0.1) | accuracy: 48.96 -> 50.34 | Improved |
| directly include all low-confidence samples (LPL-ALL vs LPL) (ablation) | FixMatch-LPL · SVHN (alpha=0.1) | accuracy: 89.68 -> 83.99 | Hurt |
| define indecisive set by global prior instead of Top-5 (ablation) | ICPL · CIFAR-100 (alpha=0.1) | accuracy: 56.58 -> 57.21 | Improved |
| define indecisive set by global prior instead of Top-1 (ablation) | ICPL · CIFAR-100 (alpha=0.1) | accuracy: 55.66 -> 57.21 | Improved |
| use proxies instead of prototypes (PROTO) (ablation) | ProxyFL · CIFAR-100 (alpha=0.1) | accuracy: 49.56 -> 57.21 | Improved |
| ProxyFL vs SAGE convergence speed to 50% accuracy | ProxyFL · CIFAR-100 (alpha=0.1) | rounds to 50% accuracy: 267 -> 177 | Improved |
| ProxyFL vs SAGE final accuracy | ProxyFL · CIFAR-100 (alpha=0.1) | accuracy: 54.18 -> 57.50 | Improved |
Closest papers by meaning
- (FL)$^2$: Overcoming Few Labels in Federated Semi-Supervised Learning (NeurIPS 2024 · similarity 0.92)
- Combating Data Imbalances in Federated Semi-supervised Learning with Dual Regulators (AAAI 2024 · similarity 0.91)
- Mind the Gap: Confidence Discrepancy Can Guide Federated Semi-Supervised Learning Across Pseudo-Mismatch (CVPR 2025 · similarity 0.90)
- BSemiFL: Semi-supervised Federated Learning via a Bayesian Approach (ICML 2025 · similarity 0.90)
- Progressive Distribution Matching for Federated Semi-Supervised Learning (AAAI 2025 · similarity 0.90)
- Exploring One-Shot Semi-supervised Federated Learning with Pre-trained Diffusion Models (AAAI 2024 · similarity 0.89)
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.
