Research map / Label-efficient and robust learning / Recognition and long-tail
Building Reliable Explanations of Unreliable Neural Networks: Locally Smoothing Perspective of Model Interpretation
Check 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 |
|---|---|---|---|
| Use RelEx instead of SmoothGrad | Natural ResNet-50 explanations · ImageNet, ε=0.07 | R: 0.18 -> 0.94 | Helped |
| Use RelEx instead of GradCAM | Robust ResNet-50 explanations · ImageNet, ε=0.07 | R: 0.58 -> 0.78 | Helped |
| Optimize saliency maps over a batch of noisy input copies (ablation) | RelEx · Adversarial examples; clean-image explanations on natural and robust ResNet-50 | target-class retrieval rate | Helped |
| Increase the number of optimization epochs (ablation) | RelEx · Adversarial examples; clean-image explanations on natural and robust ResNet-50 | target-class retrieval rate | Helped |
| Remove the noisy-input batch (ablation) | RelEx · Adversarial examples; clean-image explanations | target-class retrieval rate | Hurt |
| Use fewer optimization epochs (ablation) | RelEx · Adversarial examples; clean-image explanations | target-class retrieval rate | Hurt |
Closest papers by meaning
- Fooling Neural Network Interpretations via Adversarial Model Manipulation (NeurIPS 2019 · similarity 0.90)
- On the explainable properties of 1-Lipschitz Neural Networks: An Optimal Transport Perspective (NeurIPS 2023 · similarity 0.88)
- Robust Models Are More Interpretable Because Attributions Look Normal (ICML 2022 · similarity 0.87)
- Explaining, Fast and Slow: Abstraction and Refinement of Provable Explanations (ICML 2025 · similarity 0.87)
- Explainable Neural Networks with Guarantee: A Sparse Estimation Approach (AAAI 2025 · similarity 0.86)
- Unlearning-based Neural Interpretations (ICLR 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.