Research map / Vision-language and multimodal / Human and embodied modeling
Escaping Low-Rank Traps: Interpretable Visual Concept Learning via Implicit Vector Quantization
Shujian Gao, Yuan Wang, Chenglong Ma, Xin Gao, Jiangtao Yan, Junzhi Ning, Cheng Tang, Changkai Ji, Huihui Xu, Wei Li, Ziyan Huang, Jiashi Lin and 7 more
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 IVQ regularizer to Magnet-attention CBM (ablation) | IVQ-CBM · IDRID | BMAC (%): +11.81 BMAC (45.27 -> 73.06) | Improved |
| add IVQ regularizer (ablation) | IVQ-CBM · ISIC / NCT / BUSI / Retina accuracy | accuracy (%): ISIC 80.88 -> 90.11; IDRID 57.14 -> 67.35 | Improved |
| add Magnet attention instead of [CLS] token (ablation) | IVQ-CBM · medical datasets | accuracy (%): ISIC 89.42 -> 90.11; NCT 96.23 -> 99.90 | Improved |
| use implicit quantization instead of explicit hard VQ (ablation) | IVQ-CBM · ISIC | classification BMAC (%): 51.47 -> 86.22 | Improved |
| use implicit quantization instead of explicit VQ (ablation) | IVQ-CBM · IDRID | concept BMAC (%): 50.85 -> 59.48 | Improved |
| use IVQ instead of Spectral Regularization (ablation) | IVQ-CBM · ISIC | accuracy (%): 90.21 -> 90.11 | No clear effect |
| use IVQ instead of Barlow Twins / Gram Loss / Spectral (ablation) | IVQ-CBM · SIIM, BUSI, ISIC | BMAC (%): BUSI BMAC 91.98 -> 95.38 | Improved |
| raise feature rank via Barlow Twins or Spectral regularization beyond IVQ (ablation) | IVQ-CBM · ISIC, BUSI | accuracy / BMAC | No clear effect |
| set codebook size equal to number of text concepts (M=K) vs oversized (ablation) | IVQ-CBM · medical datasets average | ACC and BMAC | Improved |
| use a single shared prototype (M=1) (ablation) | IVQ-CBM · medical datasets average | ACC and BMAC | Hurt |
| add IVQ to counter rank collapse during training (ablation) | CBM visual features · ISIC, ImageNet feature rank | feature rank | Improved |
| replace best explainable baseline with IVQ-CBM | CBM · 13 benchmarks | accuracy (%): e.g. ImageNet-1K 72.29 -> 73.42; CIFAR-100 65.48 -> 67.12 | Improved |
Closest papers by meaning
- V2C-CBM: Building Concept Bottlenecks with Vision-to-Concept Tokenizer (AAAI 2025 · similarity 0.90)
- Discovering Fine-Grained Visual-Concept Relations by Disentangled Optimal Transport Concept Bottleneck Models (CVPR 2025 · similarity 0.88)
- Show and Tell: Visually Explainable Deep Neural Nets via Spatially-Aware Concept Bottleneck Models (CVPR 2025 · similarity 0.88)
- Hybrid Concept Bottleneck Models (CVPR 2025 · similarity 0.88)
- VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance (NeurIPS 2024 · similarity 0.87)
- Coarse-to-Fine Concept Bottleneck Models (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.
