Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-supervised Concept Bottleneck Models
Lijie Hu, Tianhao Huang, Huanyi Xie, Xilin Gong, Chenyang Ren, Zhengyu Hu, Lu Yu, Ping Ma, Di 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 alignment pseudo-label c_align and alignment loss (ablation) | SSCBM · CUB 10% labeled | concept accuracy: 80.60% -> 90.88% | Improved |
| Add alignment pseudo-label c_align and alignment loss (ablation) | SSCBM · WBCatt 10% labeled | concept accuracy: 60.49% -> 93.98% | Improved |
| Add alignment pseudo-label c_align and alignment loss (ablation) | SSCBM · 7-point 10% labeled | task accuracy: 65.82% -> 70.09% | Improved |
| Add KNN pseudo concept label c_img (ablation) | SSCBM · 7-point 10% labeled | concept accuracy: 70.06% -> 73.67% | Improved |
| Add KNN pseudo concept label c_img (ablation) | SSCBM · CUB K=1 | task accuracy: 59.89% -> 66.72% | Improved |
| Add KNN pseudo concept label c_img (ablation) | SSCBM · CUB 10% labeled | concept accuracy: 91.02% -> 90.88% | No clear effect |
| Use SSCBM instead of CEM+SSL | CEM with KNN pseudo labels · CUB K=1 | concept accuracy: 82.36% -> 88.99% | Improved |
| Use SSCBM instead of CEM+SSL | CEM with KNN pseudo labels · CUB K=1 | task accuracy: 59.35% -> 66.72% | Improved |
| Use SSCBM instead of CBM+SSL | CBM with KNN pseudo labels · CUB 10% labeled | task accuracy: 9.84% -> 67.67% | Improved |
| Increase labeled ratio from K=1 to 0.2 (ablation) | SSCBM · 7-point | concept accuracy: 66.58% -> 76.52% | Improved |
| Use 10% labeled SSCBM instead of fully supervised CEM | SSCBM · AwA2 | task accuracy: 87.00% -> 89.77% | Improved |
| Use 10% labeled SSCBM instead of fully supervised CEM | SSCBM · CUB | task accuracy: 79.82% -> 67.67% | Hurt |
Closest papers by meaning
- Hybrid Concept Bottleneck Models (CVPR 2025 · similarity 0.90)
- Energy-Based Concept Bottleneck Models: Unifying Prediction, Concept Intervention, and Probabilistic Interpretations (ICLR 2024 · similarity 0.90)
- Concepts Worth Having: Refining VLM-Guided Concept Bottleneck Models with Minimal Annotations (NeurIPS 2026 · similarity 0.89)
- VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance (NeurIPS 2024 · similarity 0.88)
- Stochastic Concept Bottleneck Models (NeurIPS 2024 · similarity 0.88)
- V2C-CBM: Building Concept Bottlenecks with Vision-to-Concept Tokenizer (AAAI 2025 · 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.
