Research map / Label-efficient and robust learning / Recognition and long-tail
Illuminating Salient Contributions in Neuron Activation With Attribution Equilibrium
Woo-Jeoung Nam, Seong–Whan Lee
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 uniform shifting Psi to the C-only rule (ablation) | AE (VGG-16) · Pointing game ablation (ALL/DIF) | pointing game: 83.9 / 69.6 -> 87.2 / 73.2 | Improved |
| add uniform shifting Psi to the C-only rule (ablation) | AE (ResNet) · Pointing game ablation (ALL/DIF) | pointing game: 85.6 / 72.6 -> 88.5 / 76.9 | Improved |
| add activated-neuron mask A and inactivated mask U to C+Psi (ablation) | AE (VGG-16) · Pointing game ablation (ALL/DIF) | pointing game: 87.2 / 73.2 -> 87.2 / 74.8 | Improved |
| add activated-neuron mask A and inactivated mask U to C+Psi (ablation) | AE (ResNet) · Pointing game ablation (ALL/DIF) | pointing game: 88.5 / 76.9 -> 88.5 / 77.1 | Improved |
| add Psi to C+A+U (ablation) | AE (VGG-16) · Pointing game ablation (ALL/DIF) | pointing game: 83.3 / 71.1 -> 87.2 / 74.8 | Improved |
| add Psi to C+A+U (ablation) | AE (ResNet) · Pointing game ablation (ALL/DIF) | pointing game: 86.4 / 74.1 -> 88.5 / 77.1 | Improved |
| add C to A+U (ablation) | AE (ResNet) · Pointing game ablation (ALL/DIF) | pointing game: 81.2 / 68.4 -> 86.4 / 74.1 | Improved |
| decrease user parameter gamma (ablation) | AE · Pointing game | pointing game | Improved |
| use AE instead of AGF/RSP for segmentation mAP | AE (VGG-16) · ImageNet segmentation | mAP: 0.598 -> 0.549 | Hurt |
| use AE instead of RSP for outside-inside ratio | AE (VGG-16) · ImageNet segmentation | outside-inside ratio (lower is better): 0.670 -> 0.456 | Improved |
| use AE instead of AGF for outside-inside ratio | AE (ResNet) · ImageNet segmentation | outside-inside ratio (lower is better): 0.767 -> 0.490 | Improved |
| use AE instead of Grad-CAM | AE (VGG-16) · PASCAL VOC 2007 pointing game (ALL, predicted) | pointing game: .945 -> .958 | Improved |
Closest papers by meaning
- Attribution Explanations for Deep Neural Networks: A Theoretical Perspective (TPAMI 2026 · similarity 0.90)
- Towards Better Visualizing the Decision Basis of Networks via Unfold and Conquer Attribution Guidance (AAAI 2023 · similarity 0.89)
- Unifying Fourteen Post-Hoc Attribution Methods With Taylor Interactions (TPAMI 2024 · similarity 0.88)
- Interpreting Deep Neural Networks with Relative Sectional Propagation by Analyzing Comparative Gradients and Hostile Activations (AAAI 2021 · similarity 0.87)
- How to Probe: Simple Yet Effective Techniques for Improving Post-hoc Explanations (ICLR 2025 · similarity 0.86)
- Neuron Activation Coverage: Rethinking Out-of-distribution Detection and Generalization (ICLR 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.
