Pith. sign in

REVIEW 4 major objections 6 minor 42 references

Fairness fixes interact non-additively: combining them across the ML pipeline often works better than any single method, but only if you systematically test the combinations.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-13 05:32 UTC pith:EPPVEXGE

load-bearing objection Solid engineering toolkit with honest negative clinical results; the multi-level non-additivity claim is real but rests on one cohort without uncertainty or code. the 4 major comments →

arxiv 2607.08953 v1 pith:EPPVEXGE submitted 2026-07-09 cs.LG

FairSelect: A Systematic Evaluation of Multi-Level and Intersectional Algorithmic Fairness

Nick Souligne , Isabella Mixton-Garcia , Vignesh Subbian This is my paper
classification cs.LG
keywords algorithmic fairnessbias mitigationintersectionalitymulti-level fairnessclinical machine learningfairness-utility tradeoffstroke risk prediction
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Most fairness methods are tested alone and on one demographic slice at a time. That leaves practitioners without guidance when real disparities cut across intersecting groups and across data preparation, model training, and decision thresholds. FairSelect is a toolkit that runs single methods and multi-stage combinations on the same models, reports fairness and accuracy for both single and intersectional subgroups, and surfaces the fairness–utility trade-offs. On synthetic clinical data built to isolate known bias mechanisms, the intended methods usually reduced the targeted gaps and multi-level combinations produced larger average fairness gains with only modest accuracy loss. On a real two-year stroke-risk prediction task, effects were highly variable: some combinations improved both fairness and performance, while many others did nothing or made disparities worse. The paper’s claim is that fairness interventions do not add up cleanly; useful configurations must be found by systematic search rather than assumed.

Core claim

Fairness interventions interact in non-additive, context-dependent ways. Targeted single methods generally reduce the disparities they were designed for under controlled conditions, yet multi-level combinations produce larger average fairness gains (and sometimes joint fairness-and-performance improvements) only when the specific mix of techniques, model, and data is evaluated systematically; many combinations remain ineffective or counterproductive.

What carries the argument

FairSelect: a modular pipeline that applies any chosen set of pre-, in-, and post-processing fairness techniques both singly and in combination, then scores the resulting models on standard performance metrics and on demographic-parity and equalized-odds differences computed over both single attributes and their intersections.

Load-bearing premise

The claim rests on synthetic datasets each engineered to isolate one bias mechanism plus a single real-world atrial-fibrillation stroke-risk cohort being enough to show that multi-level search reliably finds useful clinical fairness configurations.

What would settle it

Apply FairSelect to several additional, independent clinical prediction tasks with different outcome prevalences and demographic mixes; if multi-level combinations never outperform the best single method on both fairness and utility, or if the synthetic-to-real transfer fails, the general claim collapses.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This paper introduces FairSelect, a modular Python toolkit for systematically evaluating fairness mitigation methods applied singly and in combination across pre-, in-, and post-processing stages of supervised ML pipelines, with support for multiple model architectures and intersectional subgroup metrics. The authors validate the toolkit on synthetic clinical datasets engineered to isolate specific bias mechanisms and on a real-world replication of two-year stroke risk prediction in atrial fibrillation using All of Us EHR data. Synthetic results show that targeted methods reduce the intended disparities and that multi-level combinations yield larger average fairness gains with modest utility cost. On the clinical task, effects are highly variable: only a minority of single-method (~22–27%) and multi-level (~26%) runs improve DP_diff/EO_diff, many worsen them, while a small set of combinations improve both fairness and AUROC/accuracy. The central claim is that fairness interventions interact non-additively and context-dependently, so systematic multi-level search is needed to identify useful configurations in clinical ML.

Significance. The work addresses a genuine practical gap: fairness methods are usually evaluated in isolation and along single demographic axes, leaving little guidance for multi-stage, intersectional settings common in healthcare. The honest clinical finding that most interventions are ineffective or counterproductive is itself a useful contribution and aligns with STANDING Together-style calls for subgroup-aware evaluation. Strengths include a clear multi-stage experimental design, a controlled synthetic stress-test suite that maps bias mechanisms to techniques, support for seven model classes, and explicit reporting of both successes and failures rather than only cherry-picked wins. If the toolkit is released and the clinical evidence is statistically strengthened, FairSelect could become a useful benchmarking and selection framework for clinical ML developers. The non-additivity result is empirically grounded and of interest beyond the specific AF stroke task.

major comments (4)
  1. Section 6 / Tables 4–5: The load-bearing clinical claim that multi-level search can surface configurations that improve both fairness and utility rests on selected “best” point estimates (e.g., RF reweight→ensemble; DT reweight→ensemble→reject-option) with no standard errors, bootstrap intervals, cross-validation folds, or multiplicity correction. Given that only ~22–26% of runs improve EO_diff/DP_diff and many worsen them, it is unclear whether the reported successes are stable or sampling artifacts on a single geographic-temporal test split (N_test=2,383). Uncertainty quantification (or at least repeated splits / bootstrap) is needed for these headline configurations.
  2. Sections 4–6 and Limitations §7: External validity for the claim that FairSelect is a “practical framework … in clinical machine learning” is thin. Validation uses one AF stroke-risk cohort (All of Us) plus synthetic data designed to match the techniques’ intended mechanisms. The Limitations section acknowledges single-task scope, but the Abstract/Conclusion still generalize to clinical ML. Either narrow the claim to this task class or add at least one additional clinical prediction task (or independent cohort) so that non-additivity and “useful combination” findings are not single-dataset phenomena.
  3. Methods 3.1 and Results §§5–6: Intersectional evaluation is a stated contribution, yet clinical tables report only aggregate DP_diff and EO_diff (race and gender). It is not clear how intersectional subgroup disparities are computed, aggregated, or filtered for low-n cells (the toolkit’s filtering is mentioned only briefly in §7). Without subgroup-level or intersection-specific reporting for the clinical task, the claim that FairSelect systematically improves intersectional equity is under-supported relative to the single-axis metrics shown.
  4. Methods / Availability: Reproducibility of the toolkit and clinical pipeline is not established in the manuscript. There is no statement of code/data release for FairSelect, the synthetic generator, or the All of Us extraction scripts (beyond citing the prior AF study). For a software-and-evaluation contribution whose value is “systematic search,” public release (or a clear availability plan) is load-bearing for the practical claim.
minor comments (6)
  1. Throughout: residual line-break artifacts and spacing errors appear in the text (e.g., “approac hes”, “sou rces”, “techni ques”, “t oolkit”, “de mographic”). Clean typesetting before resubmission.
  2. Table 1: The prose states four post-processing techniques, but the visible table body lists three (Multiaccuracy Boosting, Group-Specific Youden, Reject-Option). Align the count and complete the missing row (or correct the count).
  3. Section 3.3: Synthetic generation is described at a high level; a short appendix table mapping each synthetic scenario to the intended technique and the primary fairness metric expected to move would make the stress-test design easier to audit.
  4. Table 4: Logistic Regression + Local Massaging reports AUROC (Δ) as 0.748 (0.238), which is inconsistent with the baseline AUROC of 0.773 in Table 3 (a large positive Δ would imply a large increase, not a decrease). Check sign/value of deltas for consistency.
  5. Related Work: FairLogue is cited as arXiv 2604.04858 (same group). Briefly clarify how FairSelect differs from FairLogue so readers can place the contribution relative to the authors’ prior toolkit.
  6. Section 6.3: “no combined method consistently improve both DP_diff and EO_diff” — grammar (“improve” → “improved”).

Circularity Check

0 steps flagged

No significant circularity: empirical toolkit evaluation whose claims are measured experimental outcomes, not algebraic identities or fitted parameters renamed as predictions.

full rationale

FairSelect is an empirical software/methods paper. Its load-bearing claims (non-additive multi-level interactions; some combinations improve both fairness and utility on an AF stroke-risk task) are reported as measured deltas of AUROC/ACC/DP_diff/EO_diff under baseline, single-method, and multi-level runs (Sections 5–6, Tables 3–5). Synthetic datasets are explicitly engineered as stress tests matching each technique’s intended bias mechanism (Methods 3.3); the paper treats successful correction as implementation validation, not as a first-principles prediction. No uniqueness theorem, ansatz, or fitted scale is imported to force the clinical results. The sole author-overlapping citation (FairLogue [24]) appears only in the related-work survey of evaluation tools and is not used to justify any experimental conclusion. There is therefore no self-definitional loop, no fitted-input-called-prediction, and no load-bearing self-citation chain. Score 0 is the correct honest finding.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 1 invented entities

As an empirical methods paper the load-bearing content is the experimental design and the twelve chosen techniques rather than free parameters or invented physical entities. The main assumptions are standard fairness metric definitions, the validity of sequential pre/in/post application as ‘multi-level’, and the representativeness of the synthetic bias suite and single clinical cohort.

free parameters (3)
  • Ensemble size K = K=5
    Group-balanced ensemble uses K=5 in reported best configurations; choice is not derived and affects reported fairness/utility numbers.
  • Model and technique hyperparameters
    User-specifiable but not exhaustively reported for the selected best runs; different choices can reverse fairness gains.
  • Decision thresholds (Youden, reject-option)
    Group-specific thresholds are optimized on the evaluation data; the precise optimization details and any hold-out usage are not fully specified.
axioms (3)
  • domain assumption Standard group fairness metrics (demographic parity difference, equalized odds difference, etc.) are appropriate proxies for clinical equity.
    Invoked throughout §3.1 and all results tables; the paper does not re-derive or justify them against clinical harm definitions.
  • ad hoc to paper Sequential application of one pre-, one in-, and one post-processing technique constitutes a valid multi-level intervention whose interactions can be measured.
    Defines the core experimental loop in §3; order effects and interactions among more than three techniques are left unexplored.
  • domain assumption Synthetic datasets engineered to isolate one bias mechanism are sufficient to validate that each technique corrects its intended bias.
    Stated in §3.3; used to claim implementation correctness before the clinical experiment.
invented entities (1)
  • FairSelect toolkit no independent evidence
    purpose: Unified modular pipeline that enumerates single and multi-level fairness configurations and reports intersectional metrics.
    The central software artifact; independent evidence would be public release and external reuse, which are not yet demonstrated in the paper.

pith-pipeline@v1.1.0-grok45 · 18726 in / 2859 out tokens · 48033 ms · 2026-07-13T05:32:03.002660+00:00 · methodology

0 comments
read the original abstract

Algorithmic fairness methods are increasingly used to identify and mitigate bias in machine learning models, yet most approaches are evaluated in isolation and along single demographic axes. This limits practical guidance for selecting fairness strategies, where disparities may arise across intersectional subgroups and across multiple stages of the modeling lifecycle. This work presents FairSelect, a toolkit for systematically evaluating fairness mitigation strategies applied individually and in combination across preprocessing, inprocessing, and postprocessing stages. FairSelect supports multiple model architectures, intersectional subgroup evaluation, and comparison of fairness utility tradeoffs across baseline, single method, and multi level configurations. The framework was validated using synthetic clinical datasets designed to represent specific bias mechanisms and a real-world replication of two-year stroke risk prediction among patients with atrial fibrillation. Synthetic experiments showed that targeted fairness methods generally reduced intended subgroup disparities, while combined strategies produced larger average fairness improvements with modest utility tradeoffs. In the clinical prediction task, mitigation effects were highly variable, with some combinations improving both fairness and predictive performance while others were ineffective or counterproductive. These findings demonstrate that fairness interventions interact in nonadditive and context dependent ways. FairSelect provides a practical framework for systematically identifying fairness strategies that improve subgroup equity while preserving model performance in clinical machine learning.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

42 extracted references · 8 canonical work pages

  1. [1]

    M., Harman, M., & Sarro, F

    Hort, M., Chen, Z., Zhang, J. M., Harman, M., & Sarro, F. (2023). Bias Mitigation for Machine Learning Classifiers: A Compreh ensive Survey (arXiv:2207.07068). arXiv. http://arxiv.org/abs/2207.07068

    Pith/arXiv arXiv 2023

  2. [2]

    G., Chaudhry, B., Dua, P., Goodman, K

    Walsh, C. G., Chaudhry, B., Dua, P., Goodman, K. W., Kaplan, B., Kavuluru, R., Solomonides , A., & Subbian, V. (2020). Stigma, biomarkers, and algorithmic bias: recommendations for precision behavioral health with artificial intelligence. JAMIA open, 3(1), 9 –15. https://doi.org/10.1093/jamiaopen/ooz054

    doi:10.1093/jamiaopen/ooz054 2020

  3. [3]

    L., Saseendrakumar, B

    Baxter, S. L., Saseendrakumar, B. R., Paul, P., Kim, J., Bonomi, L., Kuo, T. T., Loperena, R., Ratsimbazafy, F., Boerwinkle, E., Cicek, M., Clark, C. R., Cohn, E., Gebo, K., Mayo, K., Mockrin, S., Schully , S. D., Ramirez, A., Ohno -Machado, L., & All of Us Research Program Investigators (2021). Predictive Analytics for Glaucoma Using Data From the All of...

    doi:10.1016/j.ajo.2021.01.008 2021

  4. [4]

    M., & Purushotham, S

    Rahman, M. M., & Purushotham, S. (2022). Fair and Interpretable Models for Survival Analysis. Proceedings of the 28th ACM SIG KDD Conference on Knowledge Discovery and Data Mining, 1452–1462. https://doi.org/10.1145/3534678.3539259

    doi:10.1145/3534678.3539259 2022

  5. [5]

    F., & Beheshti, R

    Poulain, R., Bin Tarek, M. F., & Beheshti, R. (2023). Improving Fairness in AI Models on Electronic Health Records: The Case for Federated Learning Methods. Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency, 1599 –1608. https://doi.org/10.1145/3593013.3594102

    doi:10.1145/3593013.3594102 2023

  6. [6]

    E., Palmer, J., Laws, E., McCradden , M

    Alderman, J. E., Palmer, J., Laws, E., McCradden , M. D., Ordish, J., Ghassemi, M., Pfohl, S. R., Rostamzadeh, N., Cole -Lewis, H., Glocker, B., Calvert, M., Pollard, T. J., Gill, J., Gath, J., Adebajo, A., Beng, J., Leung, C. H., Kuku, S., Farmer, L.-A., … Liu, X. (2025). Tackling algorithmic bias and promoting transparency in health datasets: The STANDI...

    work page doi:10.1016/s2589-7500(24)00224-3 2025

  7. [7]

    Lin Lawrence Guo, Santiago E Arciniegas, Adam P Yan, George A Tomlinson, Melissa Beauchemin, Stephen R Pfohl, Lillian Sung, L everaging clinical epidemiology concepts to strengthen machine learning fairness evaluations, Journal of the American Medical Infor matics Association, 2026; ocag041, https://doi.org/10.1093/jamia/ocag041

    work page doi:10.1093/jamia/ocag041 2026

  8. [8]

    Dehghani, F., Paiva, P., Malik, N., Lin, J., Bayat, S., & Bento, M. (2025). Accuracy -fairness trade-off in ML for healthcare: A quantitative evaluation of bias mitigation strategies. Information and Software Technology, 188, 107896. https://doi.org/10.1016 /j.infsof.2025.107896

    arXiv 2025

  9. [9]

    A., Scheidegger, C., Venkatasubramanian, S., Choudhary, S., Hamilton, E

    Friedler, S. A., Scheidegger, C., Venkatasubramanian, S., Choudhary, S., Hamilton, E. P., & Roth, D. (2018). A comparative st udy of fairness - enhancing interventions in machine learning (arXiv:1802.04422). arXiv . https://doi.org/10.48550/arXiv.1802.04422

    Pith/arXiv arXiv doi:10.48550/arxiv.1802.04422 2018

  10. [10]

    M., Sarro, F., & Harman, M

    Chen, Z., Zhang, J. M., Sarro, F., & Harman, M. (2022). MAAT: A novel ensemble approach to addressing fairness and performance bugs for machine learning software. Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on t he Foundations of Software Engineering, 1122–1134. https://doi.org/10.1145/3540250.3549093

    doi:10.1145/3540250.3549093 2022

  11. [11]

    M., Sarro, F., & Harman, M

    Chen, Z., Zhang, J. M., Sarro, F., & Harman, M. (2023). A Comprehensive Empirical Study of Bias Mitigation Methods for Machin e Learning Classifiers. ACM Transactions on Software Engineering and Methodology, 32(4), 106:1 -106:30. https://doi.org/10.1145/3583561

    work page doi:10.1145/3583561 2023

  12. [12]

    M., Tal, I., Saber, T., Connolly, R., & Bendechache, M

    Farayola, M. M., Tal, I., Saber, T., Connolly, R., & Bendechache, M. (2026). A fairness -focused approach to recidivism prediction: Implications for accuracy, trust, and equity. AI & SOCIETY, 41(4), 2783–2801. https://doi.org/10.1007/s00146-025-02452-1

    work page doi:10.1007/s00146-025-02452-1 2026

  13. [13]

    Wastvedt, S., Huling, J., & Wolfson, J. (2023). An intersectional framework for counterfactual fairness in risk prediction. B iostatistics, kxad021. https://doi.org/10.1093/biostatistics/kxad021

    doi:10.1093/biostatistics/kxad021 2023

  14. [14]

    R., Islam, R., Keya, K

    Foulds, J. R., Islam, R., Keya, K. N., & Pan, S. (2020). An Intersectional Definition of Fairness. 2020 IEEE 36th Internation al Conference on Data Engineering (ICDE), 1918–1921. https://doi.org/10.1109/ICDE48307.2020.00203

    doi:10.1109/icde48307.2020.00203 2020

  15. [15]

    G., Lett, E., & Wan, G

    La Cava, W. G., Lett, E., & Wan, G. (2023). Fair admission risk prediction with proportional multicalibration. Proceedings of machine learning research, 209, 350–378

    2023

  16. [16]

    P., Abbasi, A., Oketch, K., Yang, Y., & Forsgren, N

    Lalor, J. P., Abbasi, A., Oketch, K., Yang, Y., & Forsgren, N. (2024). Should Fairness be a Metric or a Model? A Model-based Framework for Assessing Bias in Machine Learning Pipelines. ACM Trans. Inf. Syst., 42(4), 99:1 -99:41. https://doi.org/10.1145/3641276

    work page doi:10.1145/3641276 2024

  17. [17]

    Gao, J., Mar, P., Tang, Z.-Z., & Chen, G. (2024). Fair prediction of 2-year stroke risk in patients with atrial fibrillation. Journal of the American Medical Informatics Association, 31(12), 2820–2828. https://doi.org/10.1093/jamia/ocae170

    doi:10.1093/jamia/ocae170 2024

  18. [18]

    N., Pan, S., Sarwate, A

    Islam, R., Keya, K. N., Pan, S., Sarwate, A. D., & Foulds, J. R. (2023). Differential Fairness: An Intersectional Framework for Fair AI. Entropy, 25(4),

    2023

  19. [19]

    https://doi.org/10.3390/e25040660

    work page doi:10.3390/e25040660

  20. [20]

    Bellamy, R. K. E., Dey, K., Hind, M., Hoffman, S. C., Houde, S., Kannan, K., Lohia, P., Martino, J., Mehta, S., Mojsilovic, A ., Nagar, S., Natesan Ramamurthy, K., Richards, J., Saha, D., Sattigeri, P., Singh, M., Varshney, K. R., & Zhang, Y. (2018). AI Fai rness 360: An extensible toolkit for detecting, understanding, and mitigating unwanted algorithmic ...

    Pith/arXiv arXiv 2018

  21. [21]

    Weerts, H., Dudík, M., Edgar, R., Jalali, A., Lutz, R., & Madaio, M. (2023). Fairlearn : Assessing and improving fairness of AI systems. Journal of Machine Learning Research, 24(257), 1–8. http://jmlr.org/papers/v24/23-0389.html

    2023

  22. [22]

    Saleiro, P., Kuester, B., Stevens, A., Anisfeld, A., Hinkson, L., London, J., & Ghani, R. (2018). Aequitas: A bias and fairne ss audit toolkit. arXiv preprint arXiv:1811.05577. https://arxiv.org/abs/1811.05577

    Pith/arXiv arXiv 2018

  23. [23]

    Johnson, B., & Brun, Y. (2022). Fairkit -learn: A fairness evaluation and comparison toolkit. In Proceedings of the ACM/IEEE 44th International Conference on Software Engineering: Companion Proceedings (pp. 70 –74). Association for Computing Machinery. https://doi.org/10.1145/3510454.3516830

    doi:10.1145/3510454.3516830 2022

  24. [24]

    Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G. S., Davis, A., Dean, J., Devin, M., Ghemawat , S., Goodfellow, I., Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, Ł., Kudlur , M., Levenberg, J., Mané, D., Monga, R., Moore, S., Murray, D., Olah, C., Schuster, M., Shlens, J., Steiner, B., Sutskever, I., Tal...

    2015

  25. [25]

    Souligne, N., & Subbian, V. (2026). FairLogue: A Toolkit for Intersectional Fairness Analysis in Clinical Machine Learning Mo dels. ArXiv. https://arxiv.org/abs/2604.04858

    Pith/arXiv arXiv 2026

  26. [26]

    FairCareAI Contributors. (2026). FairCareAI: Healthcare AI fairness auditing (Version 0.2.7) [Computer software]. GitHub. htt ps://github.com/riccc- rush-lab/faircareai

    2026

  27. [27]

    Panigutti, C., Perotti, A., Panisson, A., Bajardi, P., & Pedreschi, D. (2021). FairLens: Auditing black-box clinical decision support systems. Information Processing & Management, 58(5), Article 102657. https://doi.org/10.1016/j.ipm.2021.102657

    work page doi:10.1016/j.ipm.2021.102657 2021

  28. [28]

    Wiśniewski, J., & Biecek, P. (2022). fairmodels: A flexible tool for bias detection, visualization, and mitigation in binary classification models. The R Journal, 14(1), 227–243. https://doi.org/10.32614/RJ-2022-019

    work page doi:10.32614/rj-2022-019 2022

  29. [29]

    Kozodoi, N., & Varga, T. V. (2021). fairness: Algorithmic fairness metrics (Version 1.2.1) [R package]. https://CRAN.R-project.org/package=fairness

    2021

  30. [30]

    Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2021). A Survey on Bias and Fairness in Machine Learning. ACM Comput. Surv., 54(6), 115:1-115:35. https://doi.org/10.1145/3457607

    doi:10.1145/3457607 2021

  31. [31]

    Barocas, S., Hardt, M., & Narayanan, A. (n.d.). Fairness and Machine Learning

  32. [32]

    Suresh, H., & Guttag, J. (2021). A Framework for Understanding Sources of Harm throughout the Machine Learning Life Cycle. Pr oceedings of the 1st ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization, EAAMO ’21, 1 –9. https://doi.org/10.1145/3465416.3483305

    doi:10.1145/3465416.3483305 2021

  33. [33]

    Department of Health and Human Services

    U.S. Department of Health and Human Services. (n.d.). All of Us Research Program Core Values. National Institutes of Health. https://allofus.nih.gov/article/core-values

  34. [34]

    Data preprocessing techniques for classification without discrimination

    Kamiran, F., Calders, T. Data preprocessing techniques for classification without discrimination. Knowl Inf Syst 33, 1 –33 (2012). https://doi.org/10.1007/s10115-011-0463-8

    doi:10.1007/s10115-011-0463-8 2012

  35. [35]

    V., Bowyer, K

    Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002). SMOTE: Synthetic Minority Over-sampling Technique. Journal of Artificial Intelligence Research, 16, 321–357. https://doi.org/10.1613/jair.953

    doi:10.1613/jair.953 2002

  36. [36]

    Agarwal, A., Beygelzimer, A., Dudik, M., Langford, J., & Wallach, H. (2018). A Reductions Approach to Fair Classification. Proceedings of the 35th International Conference on Machine Learning, 60–69. https://proceedings.mlr.press/v80/agarwal18a.html

    2018

  37. [37]

    Kamishima, T., Akaho, S., Asoh, H., Sakuma, J. (2012). Fairness -Aware Classifier with Prejudice Remover Regularizer. In: Flach, P.A., De Bie, T., Cristianini, N. (eds) Machine Learning and Knowledge Discovery in Databases. ECML PKDD 2012. Lecture Notes in Computer Science(), vol 75 24. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-33486-3_3

    doi:10.1007/978-3-642-33486-3_3 2012

  38. [38]

    Ursula Hebert-Johnson, Michael Kim, Omer Reingold, Guy Rothblum Proceedings of the 35th International Conference on Machine Learning, PMLR 80:1939-1948, 2018

    1939

  39. [39]

    Bagging Predictors

    Breiman, L. Bagging Predictors. Machine Learning 24, 123–140 (1996). https://doi.org/10.1023/A:1018054314350

    doi:10.1023/a:1018054314350 1996

  40. [40]

    P., Ghorbani, A., & Zou, J

    Kim, M. P., Ghorbani, A., & Zou, J. (2018). Multiaccuracy: Black -Box Post -Processing for Fairness in Classification (arXiv:1805.12317). arXiv. https://doi.org/10.48550/arXiv.1805.12317

    Pith/arXiv arXiv doi:10.48550/arxiv.1805.12317 2018

  41. [41]

    Faisal Kamiran, Asim Karim, and Xiangliang Zhang. 2012. Decision Theory for Discrimination -Aware Classification. In Proceedings of the 2012 IEEE 12th International Conference on Data Mining (ICDM '12). IEEE Computer Society, USA, 924 –929. https://doi.org/10.1109/ICDM.2012.45

    doi:10.1109/icdm.2012.45 2012

  42. [42]

    T., & Leiserson, M

    Dwork, C., Immorlica, N., Kalai, A. T., & Leiserson, M. (2018). Decoupled classifiers for group -fair and efficient machine learning. Proceedings of the 1st Conference on Fairness, Accountability and Transparency , 81, 119–133. https://proceedings.mlr.press/v81/dwork18a.html

    2018