Research map / Architectures and efficiency / CNN design
Explaining Convolutional Neural Networks through Attribution-Based Input Sampling and Block-Wise Feature Aggregation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Aggregate maps from multiple convolutional blocks with attribution-based sampling | SISE · PASCAL VOC 2007 test, ResNet-50 | EBPG: 63.24 -> 66.08 | Helped |
| Aggregate maps from multiple convolutional blocks with attribution-based sampling | SISE · PASCAL VOC 2007 test, ResNet-50 | Drop%: 35.36 -> 30.92 | Helped |
| Use attribution masks rather than randomized masks | SISE versus RISE · PASCAL VOC 2007 test, ResNet-50 | Drop%: 39.77 -> 30.92 | Helped |
| Raise the feature-map threshold μ from 0 to 0.3 (ablation) | SISE · PASCAL VOC 2007 test, ResNet-50 | Runtime (s): 9.21 -> 2.18 | Helped |
| Raise the feature-map threshold μ from 0 to 0.3 (ablation) | SISE · PASCAL VOC 2007 test, ResNet-50 | EBPG: 66.08 -> 66.54 | No clear effect |
| Raise the feature-map threshold μ from 0 to 0.5 (ablation) | SISE · PASCAL VOC 2007 test, ResNet-50 | EBPG: 66.08 -> 65.84 | Hurt |
| Raise the feature-map threshold μ from 0 to 0.5 (ablation) | SISE · PASCAL VOC 2007 test, ResNet-50 | Runtime (s): 9.21 -> 0.65 | Helped |
| Raise the feature-map threshold μ from 0 to 0.75 (ablation) | SISE · PASCAL VOC 2007 test, ResNet-50 | EBPG: 66.08 -> 62.5 | Hurt |
| Raise the feature-map threshold μ from 0 to 0.75 (ablation) | SISE · PASCAL VOC 2007 test, ResNet-50 | Runtime (s): 9.21 -> 0.38 | Helped |
Closest papers by meaning
- Interpretable and Fine-Grained Visual Explanations for Convolutional Neural Networks (CVPR 2019 · similarity 0.90)
- How to Probe: Simple Yet Effective Techniques for Improving Post-hoc Explanations (ICLR 2025 · similarity 0.89)
- Comparing the Decision-Making Mechanisms by Transformers and CNNs via Explanation Methods (CVPR 2024 · similarity 0.87)
- Keep the Faith: Faithful Explanations in Convolutional Neural Networks for Case-Based Reasoning (AAAI 2024 · similarity 0.86)
- On Gradient-like Explanation under a Black-box Setting: When Black-box Explanations Become as Good as White-box (ICML 2024 · similarity 0.86)
- DiffCAM: Data-Driven Saliency Maps by Capturing Feature Differences (CVPR 2025 · similarity 0.86)
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.