Research map / Federated learning and privacy / Personalized and heterogeneous FL
FedMPT: Federated Multi-Label Prompt Tuning of Vision-Language Models
Xucong Wang, Pengkun Wang, Zhe Zhao, Liheng Yu, Shuang Wang, Yang Wang
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 condition prompts to adapters (ablation) | Adapters only · VOC2007 | mAP: 84.40 -> 87.08 | Improved |
| add adapters to condition prompts (ablation) | CPs only · VOC2007 | mAP: 87.08 -> 87.62 | Improved |
| add optimal transport to CPs + adapters (ablation) | CPs + Adapters · VOC2007 | mAP: 87.62 -> 89.64 | Improved |
| add optimal transport to CPs only (ablation) | CPs only · VOC2007 | mAP: 87.08 -> 89.35 | Improved |
| add gating without OT (ablation) | CPs + Adapters · VOC2007 | mAP: 87.62 -> 87.89 | Improved |
| add gating with OT (ablation) | CPs + Adapters + OT · VOC2007 | mAP: 89.64 -> 90.10 | Improved |
| add four LLM-generated conditions (ablation) | FedMPT · VOC2007 | mAP: 86.48 -> 90.10 | Improved |
| increase requested conditions from 1 to 4 (ablation) | FedMPT · VOC2007 | mAP: 87.14 -> 90.10 | Improved |
| increase requested conditions from 4 to 12 (ablation) | FedMPT · VOC2007 | mAP: 90.10 -> 89.13 | Hurt |
| use vague conditions (pattern, anchor, action, habits) instead of representative ones (ablation) | FedMPT · VOC2007 | mAP: 90.10 -> 88.59 | Hurt |
| reorder the four conditions (ablation) | FedMPT · VOC2007 | mAP: 90.10 -> 89.89 | No clear effect |
| use FedMPT instead of FedMVP | FedMPT · VOC2007 (heterogeneity, avg over t) | mAP: 85.59 -> 89.51 | Improved |
Closest papers by meaning
- Language-Guided Transformer for Federated Multi-Label Classification (AAAI 2024 · similarity 0.89)
- F^3OCUS - Federated Finetuning of Vision-Language Foundation Models with Optimal Client Layer Updating Strategy via Multi-objective Meta-Heuristics (CVPR 2025 · similarity 0.87)
- Federated Learning from Vision-Language Foundation Models: Theoretical Analysis and Method (NeurIPS 2024 · similarity 0.86)
- Global and Local Prompts Cooperation via Optimal Transport for Federated Learning (CVPR 2024 · similarity 0.86)
- Correlative and Discriminative Label Grouping for Multi-Label Visual Prompt Tuning (CVPR 2025 · similarity 0.86)
- Federated Disentangled Tuning with Textual Prior Decoupling and Visual Dynamic Adaptation (ICML 2025 · 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.
