Research map / Federated learning and privacy / Personalized and heterogeneous FL
Calibrating Biased Distribution in VFM-Derived Latent Space via Cross-Domain Geometric Consistency
Yanbiao Ma, Wei Dai, Bowei Liu, Jiayi Chen, Wenke Huang, Guancheng Wan, Zhiwu Lu, Junchi Yan
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 GGEUR distribution calibration | FedAvg (CLIP+MLP) · CIFAR-100 label skew beta=0.1 | accuracy: 68.22 -> 77.70 | Improved |
| add GGEUR distribution calibration | FedAvg (CLIP+MLP) · CIFAR-100 label skew beta=0.01 | accuracy: 58.71 -> 75.72 | Improved |
| add GGEUR distribution calibration | FedAvg (CLIP+MLP) · CIFAR-100 label skew beta=0.5 | accuracy: 81.41 -> 83.31 | Improved |
| add GGEUR distribution calibration | FedAvg (CLIP+MLP) · Tiny-ImageNet label skew beta=0.01 | accuracy: 53.03 -> 64.27 | Improved |
| add GGEUR distribution calibration | FedAvg (CLIP+MLP) · CIFAR-10 label skew beta=0.01 | accuracy: 90.87 -> 94.39 | Improved |
| add GGEUR Step 1 (local augmentation) to FedAvg (ablation) | FedAvg (CLIP+MLP) · Digits | average accuracy: 76.60 -> 80.18 | Improved |
| add GGEUR Step 2 on top of Step 1 (ablation) | FedAvg (CLIP+MLP) · Digits | average accuracy: 80.18 -> 82.36 | Improved |
| add GGEUR Step 2 on top of Step 1 (ablation) | MOON (CLIP+MLP) · Digits | average accuracy: 67.50 -> 78.13 | Improved |
| add GGEUR Step 1 to MOON (ablation) | MOON (CLIP+MLP) · Digits | average accuracy: 60.07 -> 67.50 | Improved |
| add GGEUR | MOON (CLIP+MLP) · Digits | STD across domains: 15.14 -> 16.08 | Mixed |
| add GGEUR | MOON (CLIP+MLP) · Office-Home-LDS | average accuracy: 76.83 -> 78.08 | Improved |
| add GGEUR | FedOPT (CLIP+MLP) · Office-Home-LDS | average accuracy: 65.59 -> 84.20 | Improved |
Closest papers by meaning
- FedMC: Federated Manifold Calibration (ICLR 2026 · similarity 0.87)
- Geometric Prior Guided Feature Representation Learning for Long-Tailed Classification (IJCV 2024 · similarity 0.85)
- Optimizing Distributional Geometry Alignment with Optimal Transport for Generative Dataset Distillation (NeurIPS 2025 · similarity 0.85)
- Geometric Knowledge-Guided Localized Global Distribution Alignment for Federated Learning (CVPR 2025 · similarity 0.85)
- GeSS: Benchmarking Geometric Deep Learning under Scientific Applications with Distribution Shifts (NeurIPS 2024 · similarity 0.83)
- Expressive Score-Based Priors for Distribution Matching with Geometry-Preserving Regularization (ICML 2025 · similarity 0.83)
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.
