Research map / Label-efficient and robust learning / Recognition and long-tail
On the Variability of Concept Activation Vectors
Julia Wenkmann, Damien Garreau
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 |
|---|---|---|---|
| increase number of random examples N | CAV (logistic regression) · ImageNet, ResNet-50 | CAV variability | Improved |
| increase N for TCAV scores | TCAV score · ImageNet, ResNet-50 | TCAV score variance | No clear effect |
| average over s independent runs with fixed total random samples | multi-run TCAV · ImageNet, ResNet-50 | TCAV score variance | Improved |
| use Difference of Means instead of logistic regression | CAV estimator · UCI Adult tabular | variability at fixed compute | Improved |
| use Difference of Means instead of logistic regression | CAV estimator · compute time | wall-clock time: about 46x faster | Improved |
| use Concept Activation Regions instead of CAVs | CAR estimator · compute time | wall-clock time | Hurt |
| vary layer, normalization, regularization, initialization, optimizer (ablation) | CAV · all modalities | variance decay trend | No clear effect |
| change model architecture or modality (GoogLeNet, ViT, BERT, CLIP, tabular) | CAV · four modalities | variance scaling | No clear effect |
Closest papers by meaning
- FastCAV: Efficient Computation of Concept Activation Vectors for Explaining Deep Neural Networks (ICML 2025 · similarity 0.88)
- Navigating Neural Space: Revisiting Concept Activation Vectors to Overcome Directional Divergence (ICLR 2025 · similarity 0.88)
- LG-CAV: Train Any Concept Activation Vector with Language Guidance (NeurIPS 2024 · similarity 0.85)
- Sample-efficient Learning of Concepts with Theoretical Guarantees: from Data to Concepts without Interventions (NeurIPS 2025 · similarity 0.84)
- Enhancing Performance of Explainable AI Models with Constrained Concept Refinement (ICML 2025 · similarity 0.83)
- FaCT: Faithful Concept Traces for Explaining Neural Network Decisions (NeurIPS 2025 · similarity 0.83)
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.
