The explanation game: a formal framework for interpretable machine learning
Open accessThe DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index.
Abstract
We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions.
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
Rudin, C. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0048-x · Cited by 10,561 · cited 7×
“Rudin’s argument—expanded in considerable technical detail in a follow up paper (Semenova and Rudin 2019)—is premised on the assumption that sufficiently large Rashomon sets should include at least one interpretable model.”
InfluentialSearle, J. R. (1980) · Behavioral and Brain Sciences · doi:10.1017/s0140525x00005756 · Cited by 6,954
17. A Value for n-Person Games
Shapley, L. S. (1953) · Contributions to the Theory of Games (AM-28), Volume II · doi:10.1515/9781400881970-018 · Cited by 6,950
Lipton, Z. C. (2018) · Queue · doi:10.1145/3236386.3241340 · Cited by 5,427
Explanation in artificial intelligence: Insights from the social sciences
Miller, T. (2019) · Artificial Intelligence · doi:10.1016/j.artint.2018.07.007 · Cited by 5,302
Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., Luetge, C., Madelin, R., Pagallo, U., Rossi, F., et al. (2018) · Minds and Machines · doi:10.1007/s11023-018-9482-5 · Cited by 4,774 · cited 1×
“Some have argued for the creation of independent third-party groups tasked with the responsibility of auditing code under non-disclosure agreements (Floridi et al. 2018; Wachter et al. 2017), a proposal we personally support.”
Self-citationFraassen, B. C. V. (1980) · doi:10.1093/0198244274.001.0001 · Cited by 4,682
The ethics of algorithms: Mapping the debate
Mittelstadt, B. D., Allo, P., Taddeo, M., Wachter, S., & Floridi, L. (2016) · Big Data & Society · doi:10.1177/2053951716679679 · Cited by 4,567
Self-citationCausation, Prediction, and Search
Spirtes, P., Glymour, C., & Scheines, R. (2001) · doi:10.7551/mitpress/1754.001.0001 · Cited by 4,521
Statistical 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 1×
“Breiman (2001) famously introduced the idea of a Rashomon set8—a collection of models that estimate the same functional relationship using different algorithms and/or hyperparameters, yet all perform reasonably well (say, within 5% of the top performing model).”
The Black Box Society: The Secret Algorithms That Control Money and Information
Pasquale, F. (2015) · doi:10.4159/harvard.9780674736061 · Cited by 3,796
A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I., Lai, M., Guez, A., Lanctot, M., Sifre, L., Kumaran, D., Graepel, T., et al. (2018) · Science · doi:10.1126/science.aar6404 · Cited by 3,562 · cited 1×
“games like chess and Go (Silver et al., 2018), diagnosing skin cancer (Esteva et al.”
Towards A Rigorous Science of Interpretable Machine Learning
Doshi-Velez, F., & Kim, B. (2017) · arXiv · doi:10.48550/arxiv.1702.08608 · Cited by 3,121
Barocas, S., & Selbst, A. D. (2014) · doi:10.2139/ssrn.2477899 · Cited by 3,048 · cited 1×
“Any attempt to do so, however, will require the technical ability to audit algorithms in order to rigorously test whether they discriminate on the basis of protected attributes such as race and gender (Barocas and Selbst 2016).”
On the Uniform Convergence of Relative Frequencies of Events to Their Probabilities
Vapnik, V. N., & Chervonenkis, A. Y. (1971) · Theory of Probability & Its Applications · doi:10.1137/1116025 · Cited by 2,448
Causal diagrams for empirical research
PEARL, J. (1995) · Biometrika · doi:10.1093/biomet/82.4.669 · Cited by 2,355 · cited 2×
“While several overlapping formulations have been proposed for such adjustments (Galles & Pearl, 1995; Pearl, 1995; Robins, 1997), we follow Tian & Pearl (2002), who provide a provably sound and complete set of causal identifiability conditions for semi-Markovian models (Huang & Valtorta, 2008).”
Anchors: High-Precision Model-Agnostic Explanations
Ribeiro, M. T., Singh, S., & Guestrin, C. (2018) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v32i1.11491 · Cited by 2,162 · cited 2×
“Several iML solutions are built on recursive partitioning (Guidotti et al., 2018; Ribeiro et al., 2018; Yang et al., 2017) – the statistical procedure that produces rule lists – and a growing number of psychological studies suggests that users find such explanations especially intelligible (Lage et al.”
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
Planning chemical syntheses with deep neural networks and symbolic AI
Segler, M. H. S., Preuss, M., & Waller, M. P. (2018) · Nature · doi:10.1038/nature25978 · Cited by 1,847
Concentration Inequalities: A Nonasymptotic Theory of Independence
Boucheron, S., Lugosi, G., & Massart, P. (2013) · doi:10.1093/acprof:oso/9780199535255.001.0001 · Cited by 1,768
A Unified Framework of Five Principles for AI in Society
Floridi, L., & Cowls, J. (2019) · Harvard Data Science Review · doi:10.1162/99608f92.8cd550d1 · Cited by 1,631
Self-citationWachter, S., Mittelstadt, B., & Floridi, L. (2017) · International Data Privacy Law · doi:10.1093/idpl/ipx005 · Cited by 1,308
Self-citationFisher, A., Rudin, C., & Dominici, F. (2018) · arXiv · doi:10.48550/arxiv.1801.01489 · Cited by 1,165
The accuracy, fairness, and limits of predicting recidivism
Dressel, J., & Farid, H. (2018) · Science Advances · doi:10.1126/sciadv.aao5580 · Cited by 1,071
Counterfactual explanations without opening the black box: automated decisions and the GDPR
Wachter, S., Mittelstadt, B., & Russell, C. (2017) · SSRN Electronic Journal · doi:10.2139/ssrn.3063289 · Cited by 1,061
Kusner, M. J., Loftus, J. R., Russell, C., & Silva, R. (2017) · arXiv · doi:10.48550/arxiv.1703.06856 · Cited by 995 · cited 1×
“What if she had been wealthier? Or older? Would a hypothetical applicant identical to the original except along the axis of wealth or age have had more luck? Several authors in the iML literature explicitly endorse such a counterfactual strategy (Kusner et al., 2017; Wachter et al., 2018).”
A primer on deep learning in genomics
Zou, J., Huss, M., Abid, A., Mohammadi, P., Torkamani, A., & Telenti, A. (2019) · Nature Genetics · doi:10.1038/s41588-018-0295-5 · Cited by 891
Mittelstadt, B., Russell, C., & Wachter, S. (2019) · Proceedings of the Conference on Fairness, Accountability, and Transparency · doi:10.1145/3287560.3287574 · Cited by 794 · cited 1×
“The importance of contrastive explanation is highlighted by several philosophers (Hitchcock 1999; Potochnik 2015; van Fraassen 1980), and has recently begun to receive attention in the iML literature as well (Miller 2019; Mittelstadt et al. 2019).”
Lessons from the Netflix prize challenge
Bell, R. M., & Koren, Y. (2007) · ACM SIGKDD Explorations Newsletter · doi:10.1145/1345448.1345465 · Cited by 727
Predictive Policing: The Role of Crime Forecasting in Law Enforcement Operations
Perry, W., McInnis, B., Price, C., Smith, S., & Hollywood, J. (2013) · doi:10.7249/rr233 · Cited by 693
DARPA's explainable artificial intelligence (XAI) program
Gunning, D. (2019) · Proceedings of the 24th International Conference on Intelligent User Interfaces · doi:10.1145/3301275.3308446 · Cited by 546
Meaningful information and the right to explanation
Selbst, A. D., & Powles, J. (2017) · International Data Privacy Law · doi:10.1093/idpl/ipx022 · Cited by 516
Causal Inference from Complex Longitudinal Data
Robins, J. M. (1997) · Lecture Notes in Statistics · doi:10.1007/978-1-4612-1842-5_4 · Cited by 510 · cited 2×
“While several overlapping formulations have been proposed for such adjustments (Galles & Pearl, 1995; Pearl, 1995; Robins, 1997), we follow Tian & Pearl (2002), who provide a provably sound and complete set of causal identifiability conditions for semi-Markovian models (Huang & Valtorta, 2008).”
Preventing Fairness Gerrymandering: Auditing and Learning for Subgroup Fairness
Kearns, M., Neel, S., Roth, A., & Wu, Z. S. (2017) · arXiv · doi:10.48550/arxiv.1711.05144 · Cited by 469
Transparency in Algorithmic and Human Decision-Making: Is There a Double Standard?
Zerilli, J., Knott, A., Maclaurin, J., & Gavaghan, C. (2019) · Philosophy & Technology · doi:10.1007/s13347-018-0330-6 · Cited by 420 · cited 1×
“Zerilli et al. (2019) argue that proponents of iML place an unreasonable burden on algorithms by demanding that they not only perform better and faster than humans, but explain why they do so as well.”
Discrimination in the Age of Algorithms
Kleinberg, J., Ludwig, J., Mullainathan, S., & Sunstein, C. R. (2018) · Journal of Legal Analysis · doi:10.1093/jla/laz001 · Cited by 401
Slave to the Algorithm? Why a Right to Explanationn is Probably Not the Remedy You are Looking for
Edwards, L., & Veale, M. (2017) · SSRN Electronic Journal · doi:10.2139/ssrn.2972855 · Cited by 244 · cited 1×
“…such a right—some commentators insist that it does (Goodman and Flaxman 2017; Selbst and Powles 2017), while others challenge this reading (Edwards and Veale 2017; Wachter et al. 2017)—there is no question that policymakers are beginning to seriously consider the social impact of ML, and…”
A general identification condition for causal effects
(2002) · eScholarship (California Digital Library) · doi:10.5555/777092.777180 · Cited by 203
The age of secrecy and unfairness in recidivism prediction
Rudin, C., Wang, C., & Coker, B. (2020) · Harvard Data Science Review · doi:10.1162/99608f92.6ed64b30 · Cited by 177
The Rhetoric and Reality of Anthropomorphism in Artificial Intelligence
Watson, D. (2019) · Minds and Machines · doi:10.1007/s11023-019-09506-6 · Cited by 148
Self-citationAn Evaluation of the Human-Interpretability of Explanation
Lage, I., Chen, E., He, J., Narayanan, M., Kim, B., Gershman, S., & Doshi-Velez, F. (2019) · arXiv · doi:10.48550/arxiv.1902.00006 · Cited by 125 · cited 1×
“, 2017) – the statistical procedure that produces rule lists – and a growing number of psychological studies suggests that users find such explanations especially intelligible (Lage et al., 2018).”
Learning Certifiably Optimal Rule Lists for Categorical Data
Angelino, E., Larus-Stone, N., Alabi, D., Seltzer, M., & Rudin, C. (2017) · arXiv · doi:10.48550/arxiv.1704.01701 · Cited by 108
The Logic of Design as a Conceptual Logic of Information
Floridi, L. (2017) · Minds and Machines · doi:10.1007/s11023-017-9438-1 · Cited by 76 · cited 2×
“Inspired by the Formal Methods literature in computer science (Boca et al. 2010), the levelist approach has been extended to conceptualise a wide array of problems in the philosophy of information (Floridi 2008a, 2011, 2017).”
Self-citation
Cited by
The Ethics of Artificial Intelligence: Principles, Challenges, and Opportunities
Floridi, L. (2023) · doi:10.1093/oso/9780198883098.001.0001 · Cited by 507
Self-citationThe ethics of algorithms: key problems and solutions
Tsamados, A., Aggarwal, N., Cowls, J., Morley, J., Roberts, H., Taddeo, M., & Floridi, L. (2022) · AI & SOCIETY · doi:10.1007/s00146-021-01154-8 · Cited by 384
Self-citationHow Explainability Contributes to Trust in AI
Ferrario, A., & Loi, M. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3533202 · Cited by 163
Conceptual challenges for interpretable machine learning
Watson, D. S. (2022) · Synthese · doi:10.1007/s11229-022-03485-5 · Cited by 87
Self-citationConceptualising fairness: three pillars for medical algorithms and health equity
Sikstrom, L., Maslej, M. M., Hui, K., Findlay, Z., Buchman, D. Z., & Hill, S. L. (2022) · BMJ Health & Care Informatics · doi:10.1136/bmjhci-2021-100459 · Cited by 76
Metadata
- Authors
- Watson, D. S., & Floridi, L.
- Venue
- Synthese
- Published
- 2020-04-03
- Volume
- 198
- Issue
- 10
- Pages
- 9211-9242
- ISSN
- 0039-7857 (ISSN-L)
- 1573-0964
- OpenAlex
- W3014632866
- Cited by
- 71
- References
- 43
- FWCI
- 5.59
- Type
- journal article
- Language
- en
- Open access
- hybrid
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.