Research map / Label-efficient and robust learning / Recognition and long-tail
ICAM: Interpretable Classification via Disentangled Representations and Feature Attribution Mapping
Cher Bass, Mariana da Silva, Carole Sudre, Petru-Daniel Tudosiu, Stephen Smith, Emma Robinson
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 the full ICAM components and losses, including the L2 loss (ablation) | ICAM · HCP 2D lesion simulations | NCC (-): 0.385 $\pm$ 0.094 -> \bf 0.435 $\pm$ 0.092 | Improved |
| add attribute classification BCE loss and rejection sampling (ablation) | ICAM-DRIT · HCP 2D lesion simulations | NCC (+): 0.265 $\pm$ 0.062 -> 0.325 $\pm$ 0.100 | Improved |
| add feature attribution map loss (ablation) | ICAM-BCE · HCP 2D lesion simulations | NCC (+): 0.325 $\pm$ 0.100 -> \bf 0.333 $\pm$ 0.082 | Improved |
| add the full ICAM losses, including L2 loss (ablation) | ICAM-FA · HCP 2D lesion simulations | NCC (-): 0.353 $\pm$ 0.104 -> \bf 0.435 $\pm$ 0.092 | Improved |
| increase the attribute latent space size from 8 to 80 (ablation) | DRIT · HCP 2D lesion simulations | NCC (-): 0.346 $\pm$ 0.080 -> 0.380 $\pm$ 0.081 | Improved |
| use ICAM architecture with DRIT losses instead of DRIT z80 (ablation) | DRIT z80 · HCP 2D lesion simulations | NCC (-): 0.380 $\pm$ 0.081 -> 0.385 $\pm$ 0.094 | No clear effect |
| use full ICAM rather than VA-GAN | ADNI paired MCI-to-AD conversions · ADNI paired MCI-to-AD conversions | NCC (-): $0.653\pm 0.142$ -> \bf 0.683 $\pm$ 0.097 | Improved |
| use full ICAM rather than Grad-CAM | ADNI paired MCI-to-AD conversions · ADNI paired MCI-to-AD conversions | NCC (+): $0.461 \pm 0.086 $ -> \bf 0.652 $\pm$ 0.083 | Improved |
| use full ICAM rather than Integrated gradients | ADNI paired MCI-to-AD conversions · ADNI paired MCI-to-AD conversions | NCC (-): $0.378 \pm 0.064 $ -> \bf 0.683 $\pm$ 0.097 | Improved |
| use full ICAM rather than Occlusion | ADNI paired MCI-to-AD conversions · ADNI paired MCI-to-AD conversions | NCC (+): $0.354 \pm 0.057 $ -> \bf 0.652 $\pm$ 0.083 | Improved |
| use full ICAM rather than ICAM-DRIT (ablation) | UK Biobank age classification · UK Biobank age classification | latent-space class separation | Improved |
| use full ICAM rather than VA-GAN | UK Biobank ageing translation · UK Biobank ageing translation | feature attribution variability | Improved |
Later papers that compared with this method
1 later accepted paper reports ICAM as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- MaskCO: Masked Generation Drives Effective Representation Learning and Exploiting for Combinatorial Optimization (ICLR 2026 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Finer-CAM: Spotting the Difference Reveals Finer Details for Visual Explanation (CVPR 2025 · similarity 0.84)
- How to Probe: Simple Yet Effective Techniques for Improving Post-hoc Explanations (ICLR 2025 · similarity 0.84)
- One Wave To Explain Them All: A Unifying Perspective On Feature Attribution (ICML 2025 · similarity 0.84)
- Benchmarking the Attribution Quality of Vision Models (NeurIPS 2024 · similarity 0.84)
- CAPE: CAM as a Probabilistic Ensemble for Enhanced DNN Interpretation (CVPR 2024 · similarity 0.84)
- DiG-IN: Diffusion Guidance for Investigating Networks - Uncovering Classifier Differences Neuron Visualisations and Visual Counterfactual Explanations (CVPR 2024 · 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.
