Research map / Vision-language and multimodal / CLIP and zero-shot
Focus Your Attention when Few-Shot Classification
Haoqing Wang, Shibo Jie, Zhi-Hong Deng
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 FORT (key-patch attention enhancement) | LoRA, DINO ViT-B/16 · CUB 20-way 1-shot | accuracy (%): 57.9 -> 62.5 | Improved |
| add FORT | LoRA, DINO ViT-B/16 · Cars 20-way 5-shot | accuracy (%): 64.3 -> 65.7 | Improved |
| add FORT | FT, Swin-T/7 iBOT · CUB 20-way 1-shot | accuracy (%): 52.7 -> 59.1 | Improved |
| add FORT | LoRA, CLIP ViT-B/16 · CUB 20-way 1-shot | accuracy (%): 86.3 -> 87.8 | Improved |
| add FORT | LoRA, DINO ViT-B/16 · Aircraft 20-way 1-shot | accuracy (%): 26.0 -> 31.3 | Improved |
| add FORT | LoRA, DINO ViT-B/16 · Pets 20-way 5-shot | accuracy (%): 89.8 -> 90.9 | Improved |
| add FORT | FT, DINO ViT-B/16 · Pets 20-way 1-shot | accuracy (%): 71.6 -> 72.6 | No clear effect |
| add FORT | LoRA, DINO ViT-B/16 · Attention-weighted importance, 1-shot support | importance score (x10^-2): 0.71 -> 1.00 | Improved |
| fine-tune with LoRA instead of keeping backbone frozen (MetaOptNet) | DINO ViT-B/16 · Cars 20-way 1-shot | accuracy (%): 24.1 -> 23.3 | Hurt |
| use only attention for position prompts instead of attention plus gradient (ablation) | FORT with LoRA, DINO · CUB 20-way 1-shot | accuracy (%): 62.5 -> 62.1 | Hurt |
| use only gradient for position prompts instead of attention plus gradient (ablation) | FORT with LoRA, DINO · CUB 20-way 1-shot | accuracy (%): 62.5 -> 60.0 | Hurt |
| use only attention for position prompts instead of attention plus gradient (ablation) | FORT with LoRA, CLIP · CUB 20-way 1-shot | accuracy (%): 87.8 -> 86.6 | Hurt |
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.
- PLOT++ — PLOT: Prompt Learning with Optimal Transport for Vision-Language Models (ICLR 2023) · better on 1 of 1 shared result
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
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.89)
- One Meta-tuned Transformer is What You Need for Few-shot Learning (ICML 2024 · similarity 0.88)
- Making Large Vision Language Models to Be Good Few-Shot Learners (AAAI 2025 · similarity 0.88)
- Strong Baselines for Parameter-Efficient Few-Shot Fine-Tuning (AAAI 2024 · similarity 0.87)
- Rethinking Few-Shot Adaptation of Vision-Language Models in Two Stages (CVPR 2025 · similarity 0.87)
- On the Surprising Effectiveness of Attention Transfer for Vision Transformers (NeurIPS 2024 · similarity 0.87)
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.
