Research map / Label-efficient and robust learning / Recognition and long-tail
SIC: Similarity-Based Interpretable Image Classification with Neural Networks
Tom Nuno Wolf, Emre Kavak, Fabian Bongratz, Christian Wachinger
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 |
|---|---|---|---|
| SIC vs black-box backbones on Pascal VOC | SIC · Pascal VOC | accuracy: +0.29% Acc | No clear effect |
| SIC vs black-box backbone on RSNA | SIC · RSNA | balanced accuracy: -0.65% Acc | No clear effect |
| SIC vs black-box backbone on Stanford Dogs | SIC · Stanford Dogs | accuracy: 87.16 -> 83.46 | Hurt |
| use SIC instead of B-cos on Pascal VOC DenseNet121 | SIC · Pascal VOC | accuracy: 96.48 -> 96.73 | Improved |
| use SIC instead of B-cos on Stanford Dogs | SIC · Stanford Dogs | accuracy: 85.54 -> 83.46 | Hurt |
| use SIC instead of NW-Head on Stanford Dogs | SIC · Stanford Dogs | accuracy: 85.16 -> 83.46 | Hurt |
| use SIC instead of ProtoPNet on Pascal VOC DenseNet121 | SIC · Pascal VOC | accuracy: 91.19 -> 96.73 | Improved |
| use SIC instead of B-cos on FunnyBirds | SIC · FunnyBirds | background independence: 0.87 -> 0.99 | Improved |
| use SIC instead of B-cos on FunnyBirds | SIC · FunnyBirds | completeness: 0.89 -> 0.97 | Improved |
| use SIC instead of BagNet on FunnyBirds | SIC · FunnyBirds | correctness: 0.76 -> 0.72 | Hurt |
| use SIC instead of BagNet on FunnyBirds | SIC · FunnyBirds | contrastivity: 0.99 -> 0.95 | Hurt |
| use SIC instead of BagNet on FunnyBirds | SIC · FunnyBirds | completeness: 0.93 -> 0.97 | Improved |
Closest papers by meaning
- Keep the Faith: Faithful Explanations in Convolutional Neural Networks for Case-Based Reasoning (AAAI 2024 · similarity 0.87)
- Explanation-based Data Augmentation for Image Classification (NeurIPS 2021 · similarity 0.87)
- Interpretable Image Classification via Non-parametric Part Prototype Learning (CVPR 2025 · similarity 0.85)
- Explainable Neural Networks with Guarantee: A Sparse Estimation Approach (AAAI 2025 · similarity 0.85)
- I-CEE: Tailoring Explanations of Image Classification Models to User Expertise (AAAI 2024 · similarity 0.85)
- A Multimodal Automated Interpretability Agent (ICML 2024 · similarity 0.85)
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.
