Research map / Label-efficient and robust learning / Recognition and long-tail
CB-SLICE: Concept-Based Interpretable Error Slice Discovery
Yael Konforti, Mateo Espinosa Zarlenga, Elaf Almahmoud, Mateja Jamnik
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 |
|---|---|---|---|
| restrict slice formation to error-prone concepts selected by ECTP instead of the full concept set (ablation) | CB-SLICE · Waterbirds (CBM Joint) | SI, Precision@10, MGF | Improved |
| add concept-label and concept-prediction auxiliary losses to the GMM objective (ablation) | CB-SLICE · MNIST-Sum (CBM Joint) | SI, Precision@10, MGF | Improved |
| use both auxiliary losses together rather than either alone (ablation) | CB-SLICE · MNIST-Sum (CBM Joint) | SI, Precision@10, MGF | Improved |
| use GMM slicing instead of a linear layer (ablation) | CB-SLICE · MNIST-Sum (CBM Seq and Joint) | auxiliary classifier accuracy: up to ~10 points | Improved |
| replace ResNet-18 encoder with DenseNet-121 (ablation) | CB-SLICE · Waterbirds | Precision@10: 0.79 vs. 0.78 for CBM + Seq and 0.82 for CBM + Joint | No clear effect |
| use CB-SLICE instead of GEORGE on CBM | slice discovery · Waterbirds | Precision@10: up to ~65% points | Improved |
| use CB-SLICE instead of GEORGE on CBM Seq | slice discovery · MetaShift | Precision@10: 0.83 vs 0.82 | No clear effect |
| increase number of slices t_g up to about 5-10 (ablation) | CB-SLICE · MNIST-Sum | Precision@10, MGF, aux accuracy: saturate at t_g ~[5, 10] | Improved |
Closest papers by meaning
- HiBug2: Efficient and Interpretable Error Slice Discovery for Comprehensive Model Debugging (ICLR 2025 · similarity 0.86)
- Learning to Intervene on Concept Bottlenecks (ICML 2024 · similarity 0.86)
- An Analysis of Concept Bottleneck Models: Measuring, Understanding, and Mitigating the Impact of Noisy Annotations (NeurIPS 2025 · similarity 0.85)
- Visual Data Diagnosis and Debiasing with Concept Graphs (NeurIPS 2024 · similarity 0.85)
- Energy-Based Concept Bottleneck Models: Unifying Prediction, Concept Intervention, and Probabilistic Interpretations (ICLR 2024 · similarity 0.85)
- VLG-CBM: Training Concept Bottleneck Models with Vision-Language Guidance (NeurIPS 2024 · similarity 0.84)
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.
