mlsanityData to 14 Oct 2026

Certifiably Robust Interpretation in Deep Learning

arXiv 1905.12105 ↗First posted 28 May 2019cs.LG stat.MLcited by 35

Not on the map. Why?

Abstract (arXiv)

Deep learning interpretation is essential to explain the reasoning behind model predictions. Understanding the robustness of interpretation methods is important especially in sensitive domains such as medical applications since interpretation results are often used in downstream tasks. Although gradient-based saliency maps are popular methods for deep learning interpretation, recent works show that they can be vulnerable to adversarial attacks. In this paper, we address this problem and provide a certifiable defense method for deep learning interpretation. We show that a sparsified version of the popular SmoothGrad method, which computes the average saliency maps over random perturbations of the input, is certifiably robust against adversarial perturbations. We obtain this result by extending recent bounds for certifiably robust smooth classifiers to the interpretation setting. Experiments on ImageNet samples validate our theory.

Key facts. Cited by 35 papers in the map (data through 14 Oct 2026). First posted on arXiv 28 May 2019.

Builds on

No reference list yet.

Built on by

Cited by year

Citing papers per year: 2021: 12; 2022: 8; 2023: 7; 2024: 6; 2025: 2.

35 papers.

Sort:

By area:

  1. 2022The Disagreement Problem in Explainable Machine Learning: A Practitioner's Perspectivecited by 159
  2. 2022Which Explanation Should I Choose? A Function Approximation Perspective to Characterizing Post Hoc Explanationscited by 77
  3. 2021Improving Deep Learning Interpretability by Saliency Guided Trainingcited by 68
  4. 2021Skew Orthogonal Convolutionscited by 67
  5. 2022Rethinking Explainability as a Dialogue: A Practitioner's Perspectivecited by 52
  6. 2021Towards the Unification and Robustness of Perturbation and Gradient Based Explanationscited by 47
  7. 2021Improved deterministic l2 robustness on CIFAR-10 and CIFAR-100cited by 47
  8. 2021Robust Machine Learning Systems: Challenges, Current Trends, Perspectives, and the Road Aheadcited by 41
  9. 2022Fairness via Explanation Quality: Evaluating Disparities in the Quality of Post hoc Explanationscited by 40
  10. 2021Improved, Deterministic Smoothing for L_1 Certified Robustnesscited by 39
  11. 2021Low Curvature Activations Reduce Overfitting in Adversarial Trainingcited by 29
  12. 2022Understanding and Enhancing Robustness of Concept-based Modelscited by 22
23 more
  1. 2023Improving the Accuracy-Robustness Trade-Off of Classifiers via Adaptive Smoothingcited by 18
  2. 2021Mutual Adversarial Training: Learning together is better than going alonecited by 17
  3. 2022Robust Probabilistic Time Series Forecastingcited by 15
  4. 2023Interpreting and Controlling Vision Foundation Models via Text Explanationscited by 11
  5. 2021Certifiably Robust Interpretation via Renyi Differential Privacycited by 10
  6. 2022Improved techniques for deterministic l2 robustnesscited by 9
  7. 2023The Lipschitz-Variance-Margin Tradeoff for Enhanced Randomized Smoothingcited by 8
  8. 2024RAW: A Robust and Agile Plug-and-Play Watermark Framework for AI-Generated Images with Provable Guaranteescited by 8
  9. 2024Structured Gradient-based Interpretations via Norm-Regularized Adversarial Trainingcited by 7
  10. 2021Defense Against Explanation Manipulationcited by 6
  11. 2022On Transfer of Adversarial Robustness from Pretraining to Downstream Taskscited by 6
  12. 2021Deep Class-Specific Affinity-Guided Convolutional Network for Multimodal Unpaired Image Segmentationcited by 5
  13. 2023MoreauGrad: Sparse and Robust Interpretation of Neural Networks via Moreau Envelopecited by 5
  14. 2025Pixel-level Certified Explanations via Randomized Smoothingcited by 4
  15. 2021Finding Representative Interpretations on Convolutional Neural Networkscited by 3
  16. 2023Rethinking Randomized Smoothing from the Perspective of Scalabilitycited by 3
  17. 2023Mixing Classifiers to Alleviate the Accuracy-Robustness Trade-Offcited by 2
  18. 2025Bridging the Theoretical Gap in Randomized Smoothingcited by 2
  19. 2023Certified Interpretability Robustness for Class Activation Mappingcited by 1
  20. 2024Weak Robust Compatibility Between Learning Algorithms and Counterfactual Explanation Generation Algorithmscited by 1
  21. 2024Integrating uncertainty quantification into randomized smoothing based robustness guaranteescited by 1
  22. 2024Gaussian Smoothing in Saliency Maps: The Stability-Fidelity Trade-Off in Neural Network Interpretabilitycited by 1
  23. 2024A Super-pixel-based Approach to the Stable Interpretation of Neural Networkscited by 0