Epistemic values in feature importance methods: Lessons from feminist epistemology
Open accessThe DOI matches its Crossref record; 24 of the 24 references it cites were found in a registry or an index.
Abstract
As the public seeks greater accountability and transparency from machine learning algorithms, the research literature on methods to explain algorithms and their outputs has rapidly expanded. Feature importance methods form a popular class of explanation methods. In this paper, we apply the lens of feminist epistemology to recent feature importance research. We investigate what epistemic values are implicitly embedded in feature importance methods and how or whether they are in conflict with feminist epistemology. We offer some suggestions on how to conduct research on explanations that respects feminist epistemic values, taking into account the importance of social context, the epistemic privileges of subjugated knowers, and adopting more interactional ways of knowing
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
Black Feminist Thought: Knowledge, Consciousness, and the Politics of Empowerment.
Brewer, R. M., & Collins, P. H. (1992) · Contemporary Sociology · doi:10.2307/2074808 · Cited by 19,445
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 2×
“This axiom either goes unnamed [5] or appears under different names in different papers, such as “completeness” [76],”
BackgroundStatistical Modeling: The Two Cultures (with comments and a rejoinder by the author)
Breiman, L. (2001) · Statistical Science · doi:10.1214/ss/1009213726 · Cited by 4,340 · cited 5×
“Replies by Brad Efron and Bruce Hoadley in [10] argued”
BackgroundInfluentialLearning Important Features Through Propagating Activation Differences
Shrikumar, A., Greenside, P., & Kundaje, A. (2017) · arXiv · doi:10.48550/arxiv.1704.02685 · Cited by 2,761
Fairness and Abstraction in Sociotechnical Systems
Selbst, A. D., Boyd, D., Friedler, S. A., Venkatasubramanian, S., & Vertesi, J. (2019) · Proceedings of the Conference on Fairness, Accountability, and Transparency · doi:10.1145/3287560.3287598 · Cited by 1,631
Consistent Individualized Feature Attribution for Tree Ensembles
Lundberg, S. M., Erion, G. G., & Lee, S.-I. (2018) · arXiv · doi:10.48550/arxiv.1802.03888 · Cited by 1,430
Fisher, A., Rudin, C., & Dominici, F. (2018) · arXiv · doi:10.48550/arxiv.1801.01489 · Cited by 1,165
Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods
Slack, D., Hilgard, S., Jia, E., Singh, S., & Lakkaraju, H. (2020) · Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3375627.3375830 · Cited by 900
Datta, A., Sen, S., & Zick, Y. (2016) · 2016 IEEE Symposium on Security and Privacy (SP) · doi:10.1109/sp.2016.42 · Cited by 728 · cited 1×
“One option is to intervene on those features, using (for instance) their marginal distributions or a fixed baseline value, as in [18, 55]; the other is to condition on the features not in”
BackgroundManipulating and Measuring Model Interpretability
Poursabzi-Sangdeh, F., Goldstein, D. G., Hofman, J. M., Vaughan, J. W., & Wallach, H. (2018) · arXiv · doi:10.48550/arxiv.1802.07810 · Cited by 711
Staying open to interpretation: engaging multiple meanings in design and evaluation
Sengers, P., & Gaver, B. (2006) · Proceedings of the 6th conference on Designing Interactive systems · doi:10.1145/1142405.1142422 · Cited by 620 · cited 1×
“in [65], the frequencies of various email categories were represented by changes in a plant’s shape.”
BackgroundKaur, H., Nori, H., Jenkins, S., Caruana, R., Wallach, H., & Wortman Vaughan, J. (2020) · Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376219 · Cited by 564 · cited 7×
“heartening that user studies on ML explanations have started appearing in major HCI conferences [36, 41].”
BackgroundInfluentialEdwards, L., & Veale, M. (2017) · doi:10.31228/osf.io/97upg · Cited by 524
Explanations as Mechanisms for Supporting Algorithmic Transparency
Rader, E., Cotter, K., & Cho, J. (2018) · Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3173574.3173677 · Cited by 388 · cited 1×
“In the case of ML explanations, the intertwining of epistemic and social goals is particularly striking, since the demand for an explanation is often not just a demand for knowledge of how the algorithm works, but also part of an ethical demand for greater accountability or transparency [22, 60].”
BackgroundAuditing black-box models for indirect influence
Adler, P., Falk, C., Friedler, S. A., Nix, T., Rybeck, G., Scheidegger, C., Smith, B., & Venkatasubramanian, S. (2018) · Knowledge and Information Systems · doi:10.1007/s10115-017-1116-3 · Cited by 280 · cited 1×
“calculating feature importance for black-box models [2, 4, 17, 28].”
Auditing Black-Box Models for Indirect Influence
Adler, P., Falk, C., Friedler, S. A., Rybeck, G., Scheidegger, C., Smith, B., & Venkatasubramanian, S. (2016) · 2016 IEEE 16th International Conference on Data Mining (ICDM) · doi:10.1109/icdm.2016.0011 · Cited by 280
The explanation game: Explaining machine learning models using shapley values
Merrick, L., & Taly, A. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-57321-8_2 · Cited by 277
Hooker, G., Mentch, L., & Zhou, S. (2021) · Statistics and Computing · doi:10.1007/s11222-021-10057-z · Cited by 264
Problems with Shapley-value-based explanations as feature importance measures
Kumar, I. E., Venkatasubramanian, S., Scheidegger, C., & Friedler, S. (2020) · arXiv · doi:10.48550/arxiv.2002.11097 · Cited by 228 · cited 11×
“The authors of SHAP recommend using interventional distributions if we want the explanation to be “true to the model”, and observational distributions if we want the explanation to be “true to the data” [14].”
BackgroundInfluentialSelf-citationHuman factors in model interpretability: Industry practices, challenges, and needs
Hong, S. R., Hullman, J., & Bertini, E. (2020) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3392878 · Cited by 219 · cited 5×
“User studies back up the idea that explanation methods are used in very different contexts, for different purposes [35, 36].”
BackgroundInfluentialToward situated interventions for algorithmic equity: lessons from the field
Katell, M., Young, M., Dailey, D., Herman, B., Guetler, V., Tam, A., Bintz, C., Raz, D., & Krafft, P. M. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372874 · Cited by 153 · cited 1×
“places at particular points in time, rather than trying to study a system or question with an abstract approach removed from social context” [40].”
BackgroundFeature reduction for neural network based text categorization
Lam, S. L. Y., & Lee, D. L. (2003) · Proceedings. 6th International Conference on Advanced Systems for Advanced Applications · doi:10.1109/dasfaa.1999.765752 · Cited by 148 · cited 1×
“Informal descriptions of the utility of feature importance often cite the high dimensionality of useful models, implying that they could be made smaller [15, 30, 45].”
BackgroundBringing the People Back In: Contesting Benchmark Machine Learning Datasets
Denton, R., Hanna, A., Amironesei, R., Smart, A., Nicole, H., & Scheuerman, M. K. (2020) · arXiv · doi:10.48550/arxiv.2007.07399 · Cited by 100 · cited 1×
“In the broader context of ML research in general, others have warned of the pitfalls of having benchmarks that determine what “success” or “progress” is [19, 23].”
BackgroundSmith, C. E., Yu, B., Srivastava, A., Halfaker, A., Terveen, L., & Zhu, H. (2020) · Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376783 · Cited by 77
Cited by
Ethical and social risks of harm from Language Models
Weidinger, L., Mellor, J., Rauh, M., Griffin, C., Uesato, J., Huang, P.-S., Cheng, M., Glaese, M., Balle, B., Kasirzadeh, A., et al. (2021) · arXiv · doi:10.48550/arxiv.2112.04359 · Cited by 477
Auditing Large Language Models: A Three-Layered Approach
Mökander, J., Schuett, J., Kirk, H. R., & Floridi, L. (2023) · doi:10.2139/ssrn.4361607 · Cited by 292
Auditing large language models: a three-layered approach
Mökander, J., Schuett, J., Kirk, H. R., & Floridi, L. (2024) · AI and Ethics · doi:10.1007/s43681-023-00289-2 · Cited by 289
Kraft, A., & Soulier, E. (2024) · The 2024 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3630106.3658981 · Cited by 25
Kumar, I. E., Hines, K. E., & Dickerson, J. P. (2022) · Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society · doi:10.1145/3514094.3534154 · Cited by 24
“This question cannot be answered without encoding implicit epistemic values [45], and remains not well definedwithin the community at large, much less in the context of a business necessity defense.”
Self-citation
Metadata
- Authors
- Hancox-Li, L., & Kumar, I. E.
- Venue
- Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency
- Publisher
- ACM
- Published
- 2021-03
- Pages
- 817-826
- OpenAlex
- W3135484384
- Cited by
- 35
- References
- 24
- FWCI
- 2.96
- Type
- conference paper
- Language
- en
- Open access
- green
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.