mlsanityData to 14 Oct 2026

Activation-Deactivation: A General Framework for Robust Post-hoc Explainable AI

arXiv 2510.01038 ↗First posted 1 Oct 2025cs.AI cs.CV cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Perturbation-based explainability methods face criticism due to their reliance on out-of-distribution mutants. This raises doubts about the quality of the explanations. In this paper, we introduce a novel forward pass paradigm, Activation-Deactivation (AD), which obviates the need for perturbation of the input. AD replaces perturbation of input features with switching off parts of the model corresponding to to the intended perturbations. We implement ConvAD, an AD approximation algorithm for CNNs. ConvAD is a drop-in mechanism that can be easily added to any trained CNN and, without any additional training, generates more robust and more transferable explanations. We provide evaluation results across multiple architectures, datasets, methods and perturbation strategies, demonstrating the superior quality of ConvAD compared to the SOTA.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 38 of its 76 references that are in the map. First posted on arXiv 1 Oct 2025.

Builds on

38 of 76 references.

References · 38

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
  3. 2017A Unified Approach to Interpreting Model Predictionscited by 6,555
  4. 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,070
  5. 2017Axiomatic Attribution for Deep Networkscited by 3,384
  6. 2013Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Mapscited by 2,371
  7. 2013Visualizing and Understanding Convolutional Networkscited by 2,290
  8. 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,515
  9. 2017Learning Important Features Through Propagating Activation Differencescited by 1,458
  10. 2021EfficientNetV2: Smaller Models and Faster Trainingcited by 1,135
  11. 2014Striving for Simplicity: The All Convolutional Netcited by 990
  12. 2020Designing Network Design Spacescited by 960
26 more
  1. 2017Grad-CAM++: Improved Visual Explanations for Deep Convolutional Networkscited by 939
  2. 2018RISE: Randomized Input Sampling for Explanation of Black-box Modelscited by 838
  3. 2017Interpretable Explanations of Black Boxes by Meaningful Perturbationcited by 612
  4. 2020Towards Faithfully Interpretable NLP Systems: How should we define and evaluate faithfulness?cited by 599
  5. 2019Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networkscited by 582
  6. 2018A Benchmark for Interpretability Methods in Deep Neural Networkscited by 476
  7. 2019Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methodscited by 443
  8. 2019On the (In)fidelity and Sensitivity for Explanationscited by 310
  9. 2019Explanations can be manipulated and geometry is to blamecited by 216
  10. 2022Diffusion Models for Counterfactual Explanationscited by 79
  11. 2021Convolutional Dynamic Alignment Networks for Interpretable Classificationscited by 54
  12. 2022Rethinking Stability for Attribution-based Explanationscited by 52
  13. 2020BayLIME: Bayesian Local Interpretable Model-Agnostic Explanationscited by 51
  14. 2019ML-LOO: Detecting Adversarial Examples with Feature Attributioncited by 48
  15. 2021Causal Sufficiency and Actual Causationcited by 29
  16. 2020audioLIME: Listenable Explanations Using Source Separationcited by 22
  17. 2024F-Fidelity: A Robust Framework for Faithfulness Evaluation of Explainable AIcited by 18
  18. 2020MeLIME: Meaningful Local Explanation for Machine Learning Modelscited by 9
  19. 2024Causal Explanations for Image Classifierscited by 9
  20. 2020GANMEX: One-vs-One Attributions Guided by GAN-based Counterfactual Explanation Baselinescited by 8
  21. 2022Towards Improved Input Masking for Convolutional Neural Networkscited by 8
  22. 2024Explaining Image Classifierscited by 7
  23. 2013Defining Explanation in Probabilistic Systemscited by 5
  24. 2024When Can You Trust Your Explanations? A Robustness Analysis on Feature Importancescited by 5
  25. 2025I Am Big, You Are Little; I Am Right, You Are Wrongcited by 3
  26. 2025SpecReX: Explainable AI for Raman Spectroscopycited by 1

Built on by

Not cited yet.