SAM: The Sensitivity of Attribution Methods to Hyperparameters
The DOI matches its Crossref record; 20 of the 20 references it cites were found in a registry or an index.
Abstract
Attribution methods can provide powerful insights into the reasons for a classifier's decision. We argue that a key desideratum of an explanation is its robustness to input hyperparameter changes that are often randomly set or empirically tuned. High sensitivity to arbitrary hyperparameter choices does not only impede reproducibility but also questions the correctness of an explanation and impairs the trust by end-users. In this paper, we provide a thorough empirical study on the sensitivity of existing attribution methods. We found an alarming trend that many methods are highly sensitive to changes in their common hyperparameters e.g. even changing a random seed can yield a different explanation! In contrast, explanations generated for robust classifiers that are trained to be invariant to pixel-wise perturbations are surprisingly more robust. Interestingly, such sensitivity is not reflected in the average explanation correctness scores over the entire dataset as commonly reported in the literature.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
scikit-image: image processing in Python
van der Walt, S., Schönberger, J. L., Nunez-Iglesias, J., Boulogne, F., Warner, J. D., Yager, N., Gouillart, E., & Yu, T. (2014) · PeerJ · doi:10.7717/peerj.453 · Cited by 6,794
Towards Deep Learning Models Resistant to Adversarial Attacks
Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2017) · arXiv · doi:10.48550/arxiv.1706.06083 · Cited by 6,602
On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation
Bach, S., Binder, A., Montavon, G., Klauschen, F., Müller, K.-R., & Samek, W. (2015) · PLOS ONE · doi:10.1371/journal.pone.0130140 · Cited by 4,524 · cited 4×
“Therefore, many approximate methods have been proposed to modify the vanilla gradients to address the aforementioned issues [48, 51, 13].”
BackgroundMethodInfluentialDeep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images
Nguyen, A., Yosinski, J., & Clune, J. (2014) · arXiv · doi:10.48550/arxiv.1412.1897 · Cited by 3,258
Self-citationDeep neural networks are easily fooled: High confidence predictions for unrecognizable images
Nguyen, A., Yosinski, J., & Clune, J. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298640 · Cited by 3,258
Self-citationTowards A Rigorous Science of Interpretable Machine Learning
Doshi-Velez, F., & Kim, B. (2017) · arXiv · doi:10.48550/arxiv.1702.08608 · Cited by 3,121
Methods for interpreting and understanding deep neural networks
Montavon, G., Samek, W., & Müller, K.-R. (2018) · Digital Signal Processing · doi:10.1016/j.dsp.2017.10.011 · Cited by 2,665
European Union Regulations on Algorithmic Decision Making and a “Right to Explanation”
Goodman, B., & Flaxman, S. (2017) · AI Magazine · doi:10.1609/aimag.v38i3.2741 · Cited by 2,131 · cited 1×
“Why did a self-driving car decide to run into a truck [29]? Why is a patient being predicted to have breast cancer [59] or to be a future criminal [2]? The explanations for such predictions made by machine learning (ML) models can impact our lives in many ways, under scientific [53, 37], social [18] or legal [24, 19]…”
BackgroundSmoothGrad: removing noise by adding noise
Smilkov, D., Thorat, N., Kim, B., Viégas, F., & Wattenberg, M. (2017) · arXiv · doi:10.48550/arxiv.1706.03825 · Cited by 1,467 · cited 10×
“Also, taking gradients as attributions faces several issues: (1) gradient images are often noisy [48] limiting their utility; (2) gradient saturation [51] i.”
BackgroundInfluentialExplaining nonlinear classification decisions with deep taylor decomposition
Montavon, G., Lapuschkin, S., Binder, A., Samek, W., & Müller, K.-R. (2017) · Pattern Recognition · doi:10.1016/j.patcog.2016.11.008 · Cited by 1,357
Sanity Checks for Saliency Maps
Adebayo, J., Gilmer, J., Muelly, M., Goodfellow, I., Hardt, M., & Kim, B. (2018) · arXiv · doi:10.48550/arxiv.1810.03292 · Cited by 1,294
Interpretable Explanations of Black Boxes by Meaningful Perturbation
Fong, R. C., & Vedaldi, A. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.371 · Cited by 1,097 · cited 9×
“The MP optimization runs for robust models converged substantially faster within only ⇠10 steps (compared to the default Niter = 300 [22]) which can be seen in both the loss plot (Fig.”
BackgroundMethodInfluentialInterpretation of Neural Networks Is Fragile
Ghorbani, A., Abid, A., & Zou, J. (2019) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v33i01.33013681 · Cited by 765
Visualizing Deep Neural Network Decisions: Prediction Difference Analysis
Zintgraf, L. M., Cohen, T. S., Adel, T., & Welling, M. (2017) · arXiv · doi:10.48550/arxiv.1702.04595 · Cited by 487
Accountability of AI Under the Law: The Role of Explanation
Doshi-Velez, F., Kortz, M., Budish, R., Bavitz, C., Gershman, S. J., O'Brien, D., Scott, K., Shieber, S., Waldo, J., Weinberger, D., et al. (2017) · doi:10.2139/ssrn.3064761 · Cited by 457 · cited 1×
“Why did a self-driving car decide to run into a truck [29]? Why is a patient being predicted to have breast cancer [59] or to be a future criminal [2]? The explanations for such predictions made by machine learning (ML) models can impact our lives in many ways, under scientific [53, 37], social [18] or legal [24, 19]…”
BackgroundExplaining Recurrent Neural Network Predictions in Sentiment Analysis
Arras, L., Montavon, G., Müller, K.-R., & Samek, W. (2017) · Proceedings of the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis · doi:10.18653/v1/w17-5221 · Cited by 361 · cited 1×
“1) have been proposed [44] and applied to a variety of domains including natural images [35], medical brain scans [25], text [12], videos [50], and speech [14].”
BackgroundOn the Robustness of Interpretability Methods
Alvarez-Melis, D., & Jaakkola, T. S. (2018) · arXiv · doi:10.48550/arxiv.1806.08049 · Cited by 310 · cited 4×
“Explanation sensitivity First, recent work has argued that some attribution methods have a problem of being highly sensitive to small pixel-wise changes in the input image [28, 10, 23].”
BackgroundResultInfluentialInterpretable and Fine-Grained Visual Explanations for Convolutional Neural Networks
Wagner, J., Köhler, J. M., Gindele, T., Hetzel, L., Wiedemer, J. T., & Behnke, S. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00931 · Cited by 144 · cited 2×
“MP [22] is a representative of a family of methods that attempt to learn an explanation via iterative optimization [56, 40, 15, 57, 54, 9].”
BackgroundUzunova, H., Ehrhardt, J., Kepp, T., & Handels, H. (2019) · Medical Imaging 2019: Image Processing · doi:10.1117/12.2511964 · Cited by 39
Cited by
Salahuddin, Z., Woodruff, H. C., Chatterjee, A., & Lambin, P. (2022) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2021.105111 · Cited by 508
Interpretable deep learning: interpretation, interpretability, trustworthiness, and beyond
Li, X., Xiong, H., Li, X., Wu, X., Zhang, X., Liu, J., Bian, J., & Dou, D. (2022) · Knowledge and Information Systems · doi:10.1007/s10115-022-01756-8 · Cited by 497
General Pitfalls of Model-Agnostic Interpretation Methods for Machine Learning Models
Molnar, C., König, G., Herbinger, J., Freiesleben, T., Dandl, S., Scholbeck, C. A., Casalicchio, G., Grosse-Wentrup, M., & Bischl, B. (2022) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-04083-2_4 · Cited by 211
Mohamed, E., Sirlantzis, K., & Howells, G. (2022) · Displays · doi:10.1016/j.displa.2022.102239 · Cited by 67
Metadata
- Authors
- Bansal, N., Agarwal, C., & Nguyen, A.
- Venue
- 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2020-06
- Pages
- 8670-8680
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W3035561175
- Cited by
- 44
- References
- 20
- FWCI
- 3.43
- Type
- conference paper
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.