Research map / Label-efficient and robust learning / Domain adaptation and generalization
Learning Optimal Prompt Ensemble for Multi-source Visual Prompt Transfer
Enming Zhang, Liwen Cao, Yanru Wu, Zhao Zijie, Yang Li
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 H-score transferability objective to weight learning (no L_align) (ablation) | HGPrompt · VTAB-1k ablation (Cifar, DTD, Pets, Euro average) | avg accuracy: 72.5 -> 78.5 | Improved |
| add L_align gradient alignment to weight learning (no H-score) (ablation) | HGPrompt · VTAB-1k ablation (Cifar, DTD, Pets, Euro average) | avg accuracy: 72.5 -> 78.1 | Improved |
| add L_align on top of H-score (ablation) | HGPrompt · VTAB-1k ablation (Cifar, DTD, Pets, Euro average) | avg accuracy: 78.5 -> 80.0 | Improved |
| add H-score on top of L_align (ablation) | HGPrompt · VTAB-1k ablation (Cifar, DTD, Pets, Euro average) | avg accuracy: 78.1 -> 80.0 | Improved |
| HGPrompt versus strongest baseline PANDA | multi-source prompt transfer · VTAB-1k (13 tasks) | avg accuracy: 58.7 -> 60.3 | Improved |
| HGPrompt versus PANDA on SVHN | HGPrompt · VTAB-1k SVHN | accuracy: 71.2 -> 71.0 | Hurt |
| HGPrompt versus PANDA on Clevr-Count | HGPrompt · VTAB-1k Clevr-Count | accuracy: 42.8 -> 49.3 | Improved |
| H-score based weights versus SPoT weights for predicting transfer | HGPrompt · Spearman correlation, 5 tasks | Spearman rho avg: 0.105 -> 0.808 | Improved |
| H-score based weights versus PANDA weights | HGPrompt · Spearman correlation, 5 tasks | Spearman rho avg: 0.569 -> 0.808 | Improved |
| increase number of source prompts from 3 to 11 (ablation) | HGPrompt · VTAB-1k DTD target | accuracy | Improved |
| vary regularization coefficient lambda (ablation) | HGPrompt · VTAB-1k | accuracy | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PANDA — PANDA: Adapting Pretrained Features for Anomaly Detection and Segmentation (CVPR 2021) · better on 12 of 14 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Facing the Elephant in the Room: Visual Prompt Tuning or Full finetuning? (ICLR 2024 · similarity 0.89)
- Learning to Learn Better Visual Prompts (AAAI 2024 · similarity 0.89)
- Revisiting the Power of Prompt for Visual Tuning (ICML 2024 · similarity 0.89)
- VIPAMIN: Visual Prompt Initialization via Embedding Selection and Subspace Expansion (NeurIPS 2025 · similarity 0.88)
- DPC: Dual-Prompt Collaboration for Tuning Vision-Language Models (CVPR 2025 · similarity 0.88)
- Consistency-guided Prompt Learning for Vision-Language Models (ICLR 2024 · 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.
