Research map / Vision-language and multimodal / CLIP and zero-shot
Adaptive Concept Discovery for Interpretable Few-Shot Text Classification
ZHENG Lifang, Hanmo Liu, Kani Chen
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 discriminative concepts Cd to prototypical concepts Cp (ablation) | StructCBM · SST2 | accuracy: 0.7507 -> 0.8029 | Improved |
| add discriminative concepts Cd to prototypical concepts Cp (ablation) | StructCBM · AGNews | accuracy: 0.7224 -> 0.7930 | Improved |
| add unconstrained refinement (ablation) | StructCBM (Cp+Cd) · SST2 | accuracy: 0.8029 -> 0.7677 | Hurt |
| add unconstrained refinement (ablation) | StructCBM (Cp+Cd) · AGNews | accuracy: 0.7930 -> 0.8141 | Improved |
| add regularization constraint to refinement (ablation) | StructCBM · SST2 | accuracy: 0.7677 -> 0.8127 | Improved |
| add embedding model training (ablation) | StructCBM · SST2 | accuracy: 0.8127 -> 0.8390 | Improved |
| add embedding model training (ablation) | StructCBM · FinaQuery | accuracy: 0.7177 -> 0.7742 | Improved |
| switch concept generator from DeepSeek to GPT-4o | StructCBM · SST-2 | accuracy: 0.8390 -> 0.8680 | Improved |
| switch concept generator from DeepSeek to Qwen | StructCBM · SST-2 | accuracy: 0.8390 -> 0.8011 | Hurt |
| switch embedding from MiniLM to Qwen-emb | StructCBM · AGNews | accuracy: 0.8121 -> 0.8689 | Improved |
| increase shots from 1 to 10 | StructCBM · AGNews | accuracy: 0.6624 -> 0.8545 | Improved |
| increase shots from 5 to 10 | StructCBM · FinQuery | accuracy: 0.7360 -> 0.7742 | Improved |
Closest papers by meaning
- Concept Bottleneck Large Language Models (ICLR 2025 · similarity 0.89)
- Hybrid Concept Bottleneck Models (CVPR 2025 · similarity 0.88)
- Bayesian Concept Bottleneck Models with LLM Priors (NeurIPS 2025 · similarity 0.87)
- Liberating Seen Classes: Boosting Few-Shot and Zero-Shot Text Classification via Anchor Generation and Classification Reframing (AAAI 2024 · similarity 0.87)
- Interpretable but Fragile? Robustness of Concept Bottlenecks under Geometric-Semantic Perturbations (NeurIPS 2026 · similarity 0.86)
- A theoretical design of concept sets: improving the predictability of concept bottleneck models (NeurIPS 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.
