Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
ProD: Prompting-To-Disentangle Domain Knowledge for Cross-Domain Few-Shot Image Classification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use prompting-to-disentangle with DG and DS prompts | CNN+Transformer baseline · CUB, 5-way 5-shot | accuracy: 72.32% to 79.19% | Helped |
| Use prompting-to-disentangle with DG and DS prompts | CNN+Transformer baseline · CARS, 5-way 5-shot | accuracy: +6.32% | Helped |
| Use prompting-to-disentangle with DG and DS prompts | CNN+Transformer baseline · Plantae, 5-way 5-shot | accuracy: +5.77% | Helped |
| Use prompting-to-disentangle with DG and DS prompts | CNN+Transformer baseline · Places, 5-way 5-shot | accuracy: +5.87% | Helped |
| Add the shared DG prompt (ablation) | Baseline · CUB, 5-way 5-shot | accuracy: 72.32 ± 0.67 -> 75.12 ± 0.69 | Helped |
| Add the on-the-fly DS prompt (ablation) | Baseline · CUB, 5-way 5-shot | accuracy: 72.32 ± 0.67 -> 74.91 ± 0.68 | Helped |
| Add both DG and DS prompts (ablation) | Baseline · CUB, 5-way 5-shot | accuracy: 72.32 ± 0.67 -> 77.63 ± 0.74 | Helped |
| Add neutralizing loss to the DG prompt (ablation) | Baseline + DG + DS · CUB, 5-way 5-shot | accuracy: +1.02% | Helped |
| Use a local rather than global classifier for DS (ablation) | ProD · CUB, 5-way 5-shot | accuracy: 77.65 ± 0.68 -> 79.19 ± 0.63 | Helped |
| Use a local rather than global classifier for DS (ablation) | Baseline + DS · CUB, 5-way 5-shot | accuracy: 73.87 ± 0.66 -> 74.91 ± 0.68 | Helped |
| Combine DG and DS tokens for inference (ablation) | ProD · CUB, 5-way 5-shot | accuracy: 79.19 ± 0.63 | Helped |
| Add feature tokens to DG+DS inference features (ablation) | ProD · CUB, 5-way 5-shot | accuracy: 79.19 ± 0.63 -> 78.04 ± 0.72 | Hurt |
Closest papers by meaning
- Domain-Controlled Prompt Learning (AAAI 2024 · similarity 0.86)
- Random Registers for Cross-Domain Few-Shot Learning (ICML 2025 · similarity 0.86)
- A Closer Look at the CLS Token for Cross-Domain Few-Shot Learning (NeurIPS 2024 · similarity 0.86)
- Revisiting Continuity of Image Tokens for Cross-domain Few-shot Learning (ICML 2025 · similarity 0.86)
- Exploring Sparse Visual Prompt for Domain Adaptive Dense Prediction (AAAI 2024 · similarity 0.86)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 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.