mlsanityData to 7 Oct 2026

AIM: Adversarial Information Masking for Faithfulness Evaluation of Saliency Maps

arXiv 2605.16905 ↗First posted 16 May 2026cs.LG cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Post-hoc saliency methods are widely used to interpret deep neural networks, but their faithfulness is difficult to evaluate reliably. Existing evaluations mask features according to saliency-induced feature ordering and measure performance degradation, but this degradation can be confounded by the masking operator: zero masking may create out-of-distribution artifacts, while interpolation-based masking may preserve residual predictive information. We propose Adversarial Information Masking (AIM), a saliency-guided adversarial feature replacement framework for evaluating both saliency-map faithfulness and masking-operator reliability. AIM replaces selected features with values from an adversarial counterpart of the input and compares degradation under complementary masking orders. We assess reliability using random-attribution bias and stability of explanation-method faithfulness rankings. Experiments on image, audio, and EEG tasks suggest that AIM reduces masking-induced bias compared with zero and interpolation-based masking, while revealing modality-dependent differences between signed and unsigned attributions.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 40 of its 70 references that are in the map. First posted on arXiv 16 May 2026.

Builds on

40 of 70 references.

References · 40

  1. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  2. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,690
  3. 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  5. 2017Axiomatic Attribution for Deep Networkscited by 3,350
  6. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,348
  7. 2013Visualizing and Understanding Convolutional Networkscited by 2,278
  8. 2017Learning Important Features Through Propagating Activation Differencescited by 1,454
  9. 2018Sanity Checks for Saliency Mapscited by 1,149
  10. 2017SmoothGrad: removing noise by adding noisecited by 1,097
  11. 2019Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networkscited by 575
  12. 2015Evaluating the visualization of what a Deep Neural Network has learnedcited by 416
28 more
  1. 2019A Survey on Explainable Artificial Intelligence (XAI): Towards Medical XAIcited by 403
  2. 2016European Union regulations on algorithmic decision-making and a "right to explanation"cited by 343
  3. 2017The (Un)reliability of saliency methodscited by 331
  4. 2019On the (In)fidelity and Sensitivity for Explanationscited by 309
  5. 2017Real Time Image Saliency for Black Box Classifierscited by 246
  6. 2019Deep learning-based electroencephalography analysis: a systematic reviewcited by 155
  7. 2020Debugging Tests for Model Explanationscited by 146
  8. 2019Sanity Checks for Saliency Metricscited by 119
  9. 2019Smooth Grad-CAM++: An Enhanced Inference Level Visualization Technique for Deep Convolutional Neural Network Modelscited by 118
  10. 2022A Consistent and Efficient Evaluation Strategy for Attribution Methodscited by 100
  11. 2022Impossibility Theorems for Feature Attributioncited by 78
  12. 2018AudioMNIST: Exploring Explainable Artificial Intelligence for Audio Analysis on a Simple Benchmarkcited by 72
  13. 2022Post hoc Explanations may be Ineffective for Detecting Unknown Spurious Correlationcited by 59
  14. 2022Evaluation of post-hoc interpretability methods in time-series classificationcited by 53
  15. 2021Do Input Gradients Highlight Discriminative Features?cited by 52
  16. 2020Evaluations and Methods for Explanation through Robustness Analysiscited by 51
  17. 2022MAtt: A Manifold Attention Network for EEG Decodingcited by 43
  18. 2022Metrics for saliency map evaluation of deep learning explanation methodscited by 40
  19. 2019On the Vulnerability of CNN Classifiers in EEG-Based BCIscited by 17
  20. 2024F-Fidelity: A Robust Framework for Faithfulness Evaluation of Explainable AIcited by 17
  21. 2025TIMING: Temporality-Aware Integrated Gradients for Time Series Explanationcited by 9
  22. 2022Towards Best Practice of Interpreting Deep Learning Models for EEG-based Brain Computer Interfacescited by 7
  23. 2022Fidelity of Interpretability Methods and Perturbation Artifacts in Neural Networkscited by 7
  24. 2021Evaluation of Interpretability for Deep Learning algorithms in EEG Emotion Recognition: A case study in Autismcited by 6
  25. 2022Toward the application of XAI methods in EEG-based systemscited by 6
  26. 2023Explainable artificial intelligence approaches for brain-computer interfaces: a review and design spacecited by 5
  27. 2024Reliable Evaluation of Attribution Maps in CNNs: A Perturbation-Based Approachcited by 4
  28. 2024Geometric Remove-and-Retrain (GOAR): Coordinate-Invariant eXplainable AI Assessmentcited by 3

Built on by

Not cited yet.