Anomalous behaviour in loss-gradient based interpretability methods
Not on the map. Why?
Abstract (arXiv)
Loss-gradients are used to interpret the decision making process of deep learning models. In this work, we evaluate loss-gradient based attribution methods by occluding parts of the input and comparing the performance of the occluded input to the original input. We observe that the occluded input has better performance than the original across the test dataset under certain conditions. Similar behaviour is observed in sound and image recognition tasks. We explore different loss-gradient attribution methods, occlusion levels and replacement values to explain the phenomenon of performance improvement under occlusion.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 11 of its 15 references that are in the map. First posted on arXiv 15 Jul 2022.
Builds on
11 of 15 references.
References · 11
- 2014Explaining and Harnessing Adversarial Examplescited by 7,287
- 2013Intriguing properties of neural networkscited by 5,114
- 2017Learning Important Features Through Propagating Activation Differencescited by 1,454
- 2018RISE: Randomized Input Sampling for Explanation of Black-box Modelscited by 825
- 2017Interpretable Explanations of Black Boxes by Meaningful Perturbationcited by 611
- 2018A Benchmark for Interpretability Methods in Deep Neural Networkscited by 467
- 2015Evaluating the visualization of what a Deep Neural Network has learnedcited by 416
- 2018Pathologies of Neural Models Make Interpretations Difficultcited by 140
- 2017Learning how to explain neural networks: PatternNet and PatternAttributioncited by 84
- 2020Overinterpretation reveals image classification model pathologiescited by 28
- 2019Bridging Adversarial Robustness and Gradient Interpretabilitycited by 22
Built on by
Not cited yet.