pith. sign in

arxiv: 2606.23515 · v1 · pith:DFYBKRKXnew · submitted 2026-06-22 · 📊 stat.ML · cs.LG

FairBED: A Bayesian Experimental Design Approach to Gathering Fairer Data

Marcel Hedman , Emily Alger , Brieuc Lehmann , Chris Holmes , Tom Rainforth This is my paper

Pith reviewed 2026-06-26 06:05 UTC · model grok-4.3

classification 📊 stat.ML cs.LG
keywords fair machine learningBayesian experimental designdata acquisitionsensitive attributesdemographic parityinformation gainfairness quantification
0
0 comments X

The pith

FairBED gathers data by maximizing information about the target while minimizing it about sensitive attributes.

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

The paper introduces FairBED to change how training data is collected rather than trying to fix models after the fact. It defines dataset fairness as being uninformative about sensitive attributes and builds Bayesian experimental design objectives around that definition. These objectives select new data points that increase knowledge of the prediction target while limiting knowledge of protected characteristics. A theoretical connection is shown to demographic parity. Experiments indicate that models trained on the resulting datasets improve the fairness-accuracy trade-off relative to random sampling or standard BED.

Core claim

FairBED formulates fairness-aware BED objectives that maximize expected information gain about the target quantity while minimizing expected information gain about sensitive attributes, establishes a theoretical link between these objectives and demographic parity, and demonstrates that predictors trained on data acquired under FairBED exhibit improved fairness-accuracy trade-offs compared with data acquired randomly or under conventional BED.

What carries the argument

Fairness-aware BED objectives that maximize expected information gain on the target while minimizing it on sensitive attributes, using the premise that fair data is uninformative about those attributes.

If this is right

  • Data acquisition itself becomes a lever for downstream model fairness.
  • The same design can be applied to any supervised learning task that involves sensitive attributes.
  • The demographic-parity link supplies a concrete way to audit the collected dataset before model training begins.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The method could be extended to sequential acquisition loops where fairness constraints are updated after each batch.
  • In settings where sensitive attributes are only partially observed, the same information-gain trade-off might still apply by treating missing labels as additional uncertainty.

Load-bearing premise

Fair datasets should be uninformative about sensitive attributes.

What would settle it

A controlled data-acquisition experiment in which models trained on FairBED-selected data fail to show a better fairness-accuracy frontier than models trained on randomly selected data of the same size.

Figures

Figures reproduced from arXiv: 2606.23515 by Brieuc Lehmann, Chris Holmes, Emily Alger, Marcel Hedman, Tom Rainforth.

Figure 1
Figure 1. Figure 1: Conditional vs. unconditional FairBED across β (static; T=10). Lower￾bound EIG. Error bars ± 1 std err (8 seeds) [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Location Finding Designs (Static, T=10, β=0.7). Source prior mean is at (3, 3). We consider the source location finding BED benchmark [29] in a two-dimensional setting, wherein we infer a hid￾den source location ψ ∈ R p from signal strength obser￾vations y. Unlike the standard BED setup, we only want to learn about one coordinate of the source location, such that ψ = (θ, ϕ), with θ the target coordinate an… view at source ↗
Figure 3
Figure 3. Figure 3: Trade-off in target (y axis) vs sensitive [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Accuracy vs DP ratio for student gradua [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Generalization across downstream models on Census Dataset with RF used during acquisition. Results are averaged over 8 seeds (400 acquired labels). 62 64 66 68 70 Target 2 Test Time Accuracy (%) 74 76 78 80 82 84 Target 1 Test Time Accuracy (%) Model mlp rf svm rbf xgb β coefficient 10.0 3.0 1.0 0.1 EPIG [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Composition of FairBED with LFR on Graduation dataset (with LFR weight Az=1.0) compared with using LFR on model trained with random data. Points towards top right preferred. We see that FairBED provides an improved pareto front compared to random acquisition while further allowing for customization in the trade-off [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Fairness–utility frontier induced by β. Comparison of static and DAD acquisition settings. Error bars denote ± 1 standard error (too small to be seen). F.1.3 Location finding ablation study In Figs. 11 and 10 we demonstrate how β informs the trade-off between information gain in θ against information gain for the sensitive attribute ϕ for static designs and policy-based DAD [29]. Increasing β encourages un… view at source ↗
Figure 10
Figure 10. Figure 10: Sensitivity to β in the Source Lo￾cation Finding experiment with a static policy (T = 10). Increasing β promotes unlearning along ϕ at the expense of information gain along θ, with ϕ quickly plateauing while θ retains sub￾stantial information even for β > 1. Errors ± 1 std err but are too small to be seen. 0 2 4 6 8 10 β Coefficient −100 −80 −60 −40 −20 0 Percentage change in EIG (%) θ param φ param [PIT… view at source ↗
Figure 12
Figure 12. Figure 12: Sensitivity to β in the Source Location Finding experiment with a static policy (T = 10) under conditional objective (8) and unconditional objective (9). Increasing β promotes unlearning along ϕ at the expense of information gain along θ. The top left of the chart is the ideal region. Axes denote the ratio of EIG in the corresponding parameter for a policy trained under FairBED vs a policy trained to targ… view at source ↗
Figure 13
Figure 13. Figure 13: Sensitivity to horizon T (static, β=0.5). Plots show lower-bound EIG estimates. Error bars ± 1 std err. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Varying β on Census. Accuracy vs DP ratio F.2 Active Learning: Census [PITH_FULL_IMAGE:figures/full_fig_p024_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Census Income active learning curves (RF evaluation). Mean test accuracy versus acquisition iteration for the target (income; left column) and the sensitive target (gender; right column), under random label sampling (blue) and FairBED-based acquisition (orange). Each row corresponds to a different β coefficient (with β = 0 recovering standard EPIG). Curves are averaged over 8 seeds with shaded regions ind… view at source ↗
Figure 16
Figure 16. Figure 16: Census. Comparing DP ratio with number of acquired labels for different acquisition strategies. Ratio of 1 is the target. β = 0 Corresponds to EPIG (Orange) and Predictive entropy (Green) baselines. 26 [PITH_FULL_IMAGE:figures/full_fig_p026_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: presents target and sensitive attribute accuracy across an increasing number of labels for varying β and for random, HFE and FairBED acquisition. We once again demonstrate that predictive accuracy plateaus for each attribute as the label budget increases. As β increases, FairBED generally maintains predictive accuracy in the target attribute and significantly reduces predictive accuracy for the sensitive … view at source ↗
Figure 18
Figure 18. Figure 18: Student Graduation. Comparing DP ratio with number of acquired labels for different acquisition strategies. Ratio of 1 is the target. β = 0 Corresponds to EPIG (Orange) and Predictive entropy (Green) baselines. 28 [PITH_FULL_IMAGE:figures/full_fig_p028_18.png] view at source ↗
read the original abstract

Frameworks for ensuring fairness in machine learning typically focus on learning fair models from existing data. But this endeavor is often undermined by biases already present in that data. We therefore look to modify the data acquisition process itself to help gather fairer data that is inherently more suitable for training fair predictors. To this end, we introduce FairBED, which provides novel formulations for quantifying the fairness of datasets themselves based on the idea that fair datasets should be uninformative about sensitive attributes. We then use this to construct practical fairness-aware Bayesian experimental design (BED) objectives that maximize expected information gain about the target quantity of interest while minimizing expected information gain about sensitive attributes. We further derive a theoretical link between FairBED and demographic parity, and show empirically that models trained on data gathered using FairBED provide improved fairness-accuracy trade-offs compared to randomly acquired data and conventional BED.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces FairBED, a Bayesian experimental design (BED) framework for acquiring fairer data. Fairness of a dataset is quantified via lack of information about sensitive attributes; this yields BED objectives that maximize expected information gain (EIG) on the target while minimizing EIG on the sensitive attribute. A theoretical connection to demographic parity is derived, and experiments claim that models trained on FairBED data achieve better fairness-accuracy trade-offs than those trained on randomly acquired data or data from conventional BED.

Significance. If the derivations and empirical results hold, the work would be significant for shifting fairness interventions upstream to the data-acquisition stage. The explicit link between information-theoretic fairness and demographic parity, together with the dual-objective BED formulation, could influence how practitioners design data-collection pipelines in domains where sensitive attributes are known at acquisition time.

major comments (2)
  1. [Abstract / §3 (formulation)] The central modeling choice—that a fair dataset must be uninformative about the sensitive attribute—directly determines both the fairness metric and the min-EIG objective. No section appears to provide a formal justification or sensitivity analysis showing that this choice is compatible with other standard fairness definitions (e.g., equalized odds or calibration) when the downstream task is classification.
  2. [Abstract / theoretical section] The claimed theoretical link to demographic parity is stated in the abstract but the derivation steps, the precise definition of demographic parity used, and the assumptions under which the link holds are not visible in the provided material. Without these, it is impossible to assess whether the link is an equivalence, an implication, or an approximation.
minor comments (2)
  1. [Abstract] The abstract refers to “practical fairness-aware BED objectives” but supplies neither the explicit functional form of the objectives nor the approximation method used to compute EIG.
  2. [Abstract] Empirical claims compare against “randomly acquired data and conventional BED,” yet the precise experimental protocol (feature spaces, acquisition budgets, sensitive-attribute distributions, fairness and accuracy metrics) is not summarized.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments. We address each major comment below and will revise the manuscript accordingly to improve clarity.

read point-by-point responses

  1. Referee: [Abstract / §3 (formulation)] The central modeling choice—that a fair dataset must be uninformative about the sensitive attribute—directly determines both the fairness metric and the min-EIG objective. No section appears to provide a formal justification or sensitivity analysis showing that this choice is compatible with other standard fairness definitions (e.g., equalized odds or calibration) when the downstream task is classification.

    Authors: Our framework centers on the information-theoretic definition of dataset fairness as lack of information about the sensitive attribute, which directly motivates the dual EIG objective in FairBED. This modeling choice is motivated by the goal of upstream intervention during data acquisition. We did not include a formal sensitivity analysis or explicit compatibility checks against equalized odds or calibration. In the revision we will add a new subsection discussing the rationale for this choice, its relation to demographic parity, and limitations with respect to other fairness notions for classification tasks. revision: yes

  2. Referee: [Abstract / theoretical section] The claimed theoretical link to demographic parity is stated in the abstract but the derivation steps, the precise definition of demographic parity used, and the assumptions under which the link holds are not visible in the provided material. Without these, it is impossible to assess whether the link is an equivalence, an implication, or an approximation.

    Authors: Section 4 derives the link, showing that under a linear model with Gaussian priors the min-EIG objective on the sensitive attribute implies demographic parity (defined as P(Ŷ=1|S=0)=P(Ŷ=1|S=1)) in expectation for the acquired data. We will expand this section to include all derivation steps, state the precise definition of demographic parity, list the assumptions explicitly, and clarify that the result is an implication under those conditions rather than an equivalence or general approximation. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The abstract and described structure define fairness as lack of information about sensitive attributes, then construct BED objectives (max EIG target, min EIG sensitive) and claim a link to demographic parity. No quoted equations or steps reduce by construction to inputs, self-citations, or fitted parameters renamed as predictions. The central empirical claim is presented as an outcome of the method rather than forced by definition. This is the normal case of an independent construction against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract does not specify any free parameters, axioms, or invented entities. A full paper review would be required to identify these.

pith-pipeline@v0.9.1-grok · 5681 in / 1059 out tokens · 30070 ms · 2026-06-26T06:05:26.199952+00:00 · methodology

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

83 extracted references · 12 canonical work pages · 1 internal anchor

  1. [1]

    UCI Machine Learning Repository, 2000

    Census-Income (KDD). UCI Machine Learning Repository, 2000. DOI: https://doi.org/10.24432/C5N30T. 8

    work page doi:10.24432/c5n30t 2000

  2. [2]

    Active sampling for min-max fairness.arXiv preprint arXiv:2006.06879, 2020

    Abernethy, J., Awasthi, P., Kleindessner, M., Morgenstern, J., Russell, C., and Zhang, J. Active sampling for min-max fairness.arXiv preprint arXiv:2006.06879, 2020. 6

    arXiv 2006

  3. [3]

    S., Sundaram, S

    Abraham, S. S., Sundaram, S. S., et al. Fairness in clustering with multiple sensitive attributes. arXiv preprint arXiv:1910.05113, 2019. 7

    arXiv 1910

  4. [4]

    A reductions approach to fair classification

    Agarwal, A., Beygelzimer, A., Dudík, M., Langford, J., and Wallach, H. A reductions approach to fair classification. InInternational conference on machine learning, pp. 60–69. PMLR, 2018. 7

    2018

  5. [5]

    Fair active learning.Expert Sys- tems with Applications, 199:116981, 2022

    Anahideh, H., Asudeh, A., and Thirumuruganathan, S. Fair active learning.Expert Sys- tems with Applications, 199:116981, 2022. ISSN 0957-4174. doi: https://doi.org/10.1016/ j.eswa.2022.116981. URL https://www.sciencedirect.com/science/article/pii/ S0957417422004055. 7

    arXiv 2022

  6. [6]

    and Li, J

    Ao, Z. and Li, J. On estimating the gradient of the expected information gain in bayesian exper- imental design. InProceedings of the AAAI Conference on Artificial Intelligence, volume 38, pp. 20311–20319, 2024. 2

    2024

  7. [7]

    T., Zhang, C., Krishnamurthy, A., Langford, J., and Agarwal, A

    Ash, J. T., Zhang, C., Krishnamurthy, A., Langford, J., and Agarwal, A. Deep batch active learning by diverse, uncertain gradient lower bounds. InInternational Conference on Learning Representations, 2020. URLhttps://openreview.net/forum?id=ryghZJBKPS. 6

    2020

  8. [8]

    Barlas, Y . Z. and Salako, K. Performance comparisons of reinforcement learning algorithms for sequential experimental design, 2025. URLhttps://arxiv.org/abs/2503.05905. 20

    arXiv 2025

  9. [9]

    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., Ramamurthy, K. N., Richards, J., Saha, D., Sattigeri, P., Singh, M., Varshney, K. R., and Zhang, Y . AI Fairness 360: An extensible toolkit for detecting, understanding, and mitigating unwanted algorithmic bias, Octobe...

    Pith/arXiv arXiv 2018

  10. [10]

    Prediction- oriented bayesian active learning

    Bickford Smith, F., Kirsch, A., Farquhar, S., Gal, Y ., Foster, A., and Rainforth, T. Prediction- oriented bayesian active learning. In Ruiz, F., Dy, J., and van de Meent, J.-W. (eds.),Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, volume 206 of Proceedings of Machine Learning Research, pp. 7331–7348. PMLR, 25–...

    2023

  11. [11]

    Making better use of unlabelled data in bayesian active learning.ArXiv, abs/2404.17249, 2024

    Bickford-Smith, F., Foster, A., and Rainforth, T. Making better use of unlabelled data in bayesian active learning.ArXiv, abs/2404.17249, 2024. URL https://api.semanticscholar.org/ CorpusID:269430415. 6

    arXiv 2024

  12. [12]

    V ., Chades, I., and Dezfouli, A

    Blau, T., Bonilla, E. V ., Chades, I., and Dezfouli, A. Optimizing sequential experimental design with deep reinforcement learning. InInternational conference on machine learning, pp. 2107–2128. PMLR, 2022. 2, 20

    2022

  13. [13]

    R., Intes, X., Bürkner, P.-C., and Radev, S

    Bracher, N., Kühmichel, L., Ivanova, D. R., Intes, X., Bürkner, P.-C., and Radev, S. T. Jadai: Jointly amortizing adaptive design and bayesian inference, 2025. URL https://arxiv.org/ abs/2512.22999. 20

    arXiv 2025

  14. [14]

    Adaptive sampling strategies to construct equitable training datasets

    Cai, W., Encarnacion, R., Chern, B., Corbett-Davies, S., Bogen, M., Bergman, S., and Goel, S. Adaptive sampling strategies to construct equitable training datasets. InProceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, FAccT ’22, pp. 1467–1478, New York, NY , USA, 2022. Association for Computing Machinery. ISBN 9781450393...

    work page doi:10.1145/3531146.3533203 2022

  15. [15]

    Adaptive sampling strategies to construct equitable training datasets

    Cai, W., Encarnacion, R., Chern, B., Corbett-Davies, S., Bogen, M., Bergman, S., and Goel, S. Adaptive sampling strategies to construct equitable training datasets. InProceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, pp. 1467–1478, 2022. 7

    2022

  16. [16]

    and Haas, C

    Caton, S. and Haas, C. Fairness in machine learning: A survey.ACM Comput. Surv., 56(7), April

  17. [17]

    doi: 10.1145/3616865

    ISSN 0360-0300. doi: 10.1145/3616865. URL https://doi.org/10.1145/3616865. 1, 3

    work page doi:10.1145/3616865

  18. [18]

    R., Myung, J

    Cavagnaro, D. R., Myung, J. I., Pitt, M. A., and Kujala, J. V . Adaptive design optimization: A mutual information-based approach to model discrimination in cognitive science.Neural computation, 22(4):887–905, 2010. 2

    2010

  19. [19]

    and Wang, X

    Chai, J. and Wang, X. Fairness with adaptive weights. In Chaudhuri, K., Jegelka, S., Song, L., Szepesvari, C., Niu, G., and Sabato, S. (eds.),Proceedings of the 39th International Conference on Machine Learning, volume 162 ofProceedings of Machine Learning Research, pp. 2853–

  20. [20]

    URL https://proceedings.mlr.press/v162/chai22a

    PMLR, 17–23 Jul 2022. URL https://proceedings.mlr.press/v162/chai22a. html. 1, 3

    2022

  21. [21]

    and Verdinelli, I

    Chaloner, K. and Verdinelli, I. Bayesian experimental design: A review.Statistical Science, pp. 273–304, 1995. 1, 2

    1995

  22. [22]

    Fair prediction with disparate impact: A study of bias in recidivism prediction instruments, 2017

    Chouldechova, A. Fair prediction with disparate impact: A study of bias in recidivism prediction instruments, 2017. URLhttps://arxiv.org/abs/1703.00056. 1, 3

    Pith/arXiv arXiv 2017

  23. [23]

    Cruz, A. F. and Hardt, M. Unprocessing seven years of algorithmic fairness.arXiv preprint arXiv:2306.07261, 2023. 7

    arXiv 2023

  24. [24]

    Review of mathematical frameworks for fairness in machine learning, 2020

    del Barrio, E., Gordaliza, P., and Loubes, J.-M. Review of mathematical frameworks for fairness in machine learning, 2020. URLhttps://arxiv.org/abs/2005.13755. 2, 3, 30

    arXiv 2020

  25. [25]

    Fairness through awareness,

    Dwork, C., Hardt, M., Pitassi, T., Reingold, O., and Zemel, R. Fairness through awareness,

  26. [26]

    URLhttps://arxiv.org/abs/1104.3913. 3, 6

    Pith/arXiv arXiv

  27. [27]

    T., and Leiserson, M

    Dwork, C., Immorlica, N., Kalai, A. T., and Leiserson, M. Decoupled classifiers for group-fair and efficient machine learning. InConference on fairness, accountability and transparency, pp. 119–133. PMLR, 2018. 6

    2018

  28. [28]

    M., Saxena, A., Pei, Y ., and Pechenizkiy, M

    Fajri, R. M., Saxena, A., Pei, Y ., and Pechenizkiy, M. Fal-cur: Fair active learning using uncertainty and representativeness on fair clustering.Expert Syst. Appl., 242(C), May 2024. ISSN 0957-4174. doi: 10.1016/j.eswa.2023.122842. URL https://doi.org/10.1016/j. eswa.2023.122842. 7

    work page doi:10.1016/j.eswa.2023.122842 2024

  29. [29]

    A., Moeller, J., Scheidegger, C., and Venkatasubramanian, S

    Feldman, M., Friedler, S. A., Moeller, J., Scheidegger, C., and Venkatasubramanian, S. Cer- tifying and removing disparate impact. InProceedings of the 21th ACM SIGKDD Interna- tional Conference on Knowledge Discovery and Data Mining, KDD ’15, pp. 259–268, New York, NY , USA, 2015. Association for Computing Machinery. ISBN 9781450336642. doi: 10.1145/2783...

    work page doi:10.1145/2783258.2783311 2015

  30. [30]

    W., Rainforth, T., and Goodman, N

    Foster, A., Jankowiak, M., Bingham, E., Horsfall, P., Teh, Y . W., Rainforth, T., and Goodman, N. Variational Bayesian Optimal Experimental Design. InAdvances in Neural Information Processing Systems 32, pp. 14036–14047. Curran Associates, Inc., 2019. 2

    2019

  31. [31]

    W., and Rainforth, T

    Foster, A., Jankowiak, M., O’Meara, M., Teh, Y . W., and Rainforth, T. A unified stochastic gradient approach to designing bayesian-optimal experiments. InInternational Conference on Artificial Intelligence and Statistics, pp. 2959–2969. PMLR, 2020. 7, 18

    2020

  32. [32]

    R., Malik, I., and Rainforth, T

    Foster, A., Ivanova, D. R., Malik, I., and Rainforth, T. Deep adaptive design: Amortizing sequential bayesian experimental design.Proceedings of the 38th International Conference on Machine Learning (ICML), PMLR 139, 2021. 2, 7, 18, 20, 21, 22

    2021

  33. [33]

    R., Thurston, H., Varghese, P., Hong, C., and Gronsbell, J

    Gao, J., Chou, B., McCaw, Z. R., Thurston, H., Varghese, P., Hong, C., and Gronsbell, J. What is fair? defining fairness in machine learning for health.Statistics in Medicine, 44(20-22): e70234, 2025. 30

    2025

  34. [34]

    Unbiased mlmc stochastic gradient-based optimization of bayesian experimental designs.arXiv preprint arXiv:2005.08414, 2020

    Goda, T., Hironaka, T., and Kitade, W. Unbiased mlmc stochastic gradient-based optimization of bayesian experimental designs.arXiv preprint arXiv:2005.08414, 2020. 2

    arXiv 2005

  35. [35]

    Equality of opportunity in super- vised learning

    Hardt, M., Price, E., Price, E., and Srebro, N. Equality of opportunity in super- vised learning. In Lee, D., Sugiyama, M., Luxburg, U., Guyon, I., and Garnett, R. (eds.),Advances in Neural Information Processing Systems, volume 29. Curran Associates, Inc., 2016. URL https://proceedings.neurips.cc/paper_files/paper/2016/file/ 9d2682367c3935defcb1f9e247a97...

    2016

  36. [36]

    Equality of opportunity in supervised learning.Advances in neural information processing systems, 29, 2016

    Hardt, M., Price, E., and Srebro, N. Equality of opportunity in supervised learning.Advances in neural information processing systems, 29, 2016. 6, 7

    2016

  37. [37]

    Deep residual learning for image recognition

    He, K., Zhang, X., Ren, S., and Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR),

  38. [38]

    and Li, Y

    He, X. and Li, Y . Fairness in machine learning: A review for statisticians.Journal of the American Statistical Association, 120(552):2834–2851, 2025. doi: 10.1080/01621459.2025. 2579579. URLhttps://doi.org/10.1080/01621459.2025.2579579. 1, 3

    work page doi:10.1080/01621459.2025 2025

  39. [39]

    R., Guan, C., and Rainforth, T

    Hedman, M., Ivanova, D. R., Guan, C., and Rainforth, T. Step-DAD: Semi-amortized policy- based bayesian experimental design. InForty-second International Conference on Machine Learning, 2025. URLhttps://openreview.net/forum?id=JRg8P2bX8P. 2, 20

    2025

  40. [40]

    Bayesian active learning for classification and preference learning.arXiv preprint arXiv:1112.5745, 2011

    Houlsby, N., Huszár, F., Ghahramani, Z., and Lengyel, M. Bayesian active learning for classification and preference learning.arXiv preprint arXiv:1112.5745, 2011. 6, 19

    Pith/arXiv arXiv 2011

  41. [41]

    Pushing the limits of fairness impossibility: Who’s the fairest of them all?Advances in Neural Information Processing Systems, 35: 32749–32761, 2022

    Hsu, B., Mazumder, R., Nandy, P., and Basu, K. Pushing the limits of fairness impossibility: Who’s the fairest of them all?Advances in Neural Information Processing Systems, 35: 32749–32761, 2022. 30

    2022

  42. [42]

    Optimal experimental design: Formulations and computations.Acta Numerica, 33:715–840, 2024

    Huan, X., Jagalur, J., and Marzouk, Y . Optimal experimental design: Formulations and computations.Acta Numerica, 33:715–840, 2024. 2

    2024

  43. [43]

    Amortized bayesian experimental design for decision-making

    Huang, D., Guo, Y ., Acerbi, L., and Kaski, S. Amortized bayesian experimental design for decision-making. InProceedings of the 38th International Conference on Neural Information Processing Systems, NIPS ’24, Red Hook, NY , USA, 2024. Curran Associates Inc. ISBN 9798331314385. 20

    2024

  44. [44]

    Aline: Joint amortization for bayesian inference and active data acquisition.ArXiv, abs/2506.07259, 2025

    Huang, D., Wen, X., Bharti, A., Kaski, S., and Acerbi, L. Aline: Joint amortization for bayesian inference and active data acquisition.ArXiv, abs/2506.07259, 2025. URL https: //api.semanticscholar.org/CorpusID:279251169. 20

    arXiv 2025

  45. [45]

    Bayesian experimental design via contrastive diffusions, 2025

    Iollo, J., Heinkelé, C., Alliez, P., and Forbes, F. Bayesian experimental design via contrastive diffusions, 2025. URLhttps://arxiv.org/abs/2410.11826. 20

    arXiv 2025

  46. [46]

    Nesting particle filters for experimental design in dynamical systems, 2024

    Iqbal, S., Corenflos, A., Särkkä, S., and Abdulsamad, H. Nesting particle filters for experimental design in dynamical systems, 2024. URLhttps://arxiv.org/abs/2402.07868. 2 12

    arXiv 2024

  47. [47]

    R., Foster, A., Kleinegesse, S., Gutmann, M., and Rainforth, T

    Ivanova, D. R., Foster, A., Kleinegesse, S., Gutmann, M., and Rainforth, T. Implicit Deep Adaptive Design: Policy–Based Experimental Design without Likelihoods. InAd- vances in Neural Information Processing Systems, volume 34, pp. 25785–25798. Curran Associates, Inc., 2021. URL https://proceedings.neurips.cc/paper/2021/file/ d811406316b669ad3d370d78b51b1d...

    2021

  48. [48]

    Decision theory for discrimination-aware classification

    Kamiran, F., Karim, A., and Zhang, X. Decision theory for discrimination-aware classification. In2012 IEEE 12th international conference on data mining, pp. 924–929. IEEE, 2012. 7, 10, 31

    2012

  49. [49]

    A., and Rainforth, T

    Kerrigan, G., Naesseth, C. A., and Rainforth, T. A geometric approach to optimal experimental design. InThe 29th International Conference on Artificial Intelligence and Statistics, 2026. URLhttps://openreview.net/forum?id=u0aepMHQ5p. 20

    2026

  50. [50]

    Kleinegesse, S., Drovandi, C., and Gutmann, M. U. Sequential Bayesian Experimental Design for Implicit Models via Mutual Information.Bayesian Analysis, pp. 1 – 30, 2021. doi: 10.1214/20-BA1225. 20

    work page doi:10.1214/20-ba1225 2021

  51. [51]

    Adaptive sensitive reweighting to mitigate bias in fairness-aware classification

    Krasanakis, E., Spyromitros-Xioufis, E., Papadopoulos, S., and Kompatsiaris, Y . Adaptive sensitive reweighting to mitigate bias in fairness-aware classification. InProceedings of the 2018 World Wide Web Conference, WWW ’18, pp. 853–862, Republic and Canton of Geneva, CHE,

    2018

  52. [52]

    ISBN 9781450356398

    International World Wide Web Conferences Steering Committee. ISBN 9781450356398. doi: 10.1145/3178876.3186133. URL https://doi.org/10.1145/3178876.3186133. 1, 3

    work page doi:10.1145/3178876.3186133

  53. [53]

    Adaptive sensitive reweighting to mitigate bias in fairness-aware classification

    Krasanakis, E., Spyromitros-Xioufis, E., Papadopoulos, S., and Kompatsiaris, Y . Adaptive sensitive reweighting to mitigate bias in fairness-aware classification. InProceedings of the 2018 world wide web conference, pp. 853–862, 2018. 6

    2018

  54. [54]

    Policy-based bayesian experimental design for non-differentiable implicit models.arXiv preprint arXiv:2203.04272,

    Lim, V ., Novoseller, E., Ichnowski, J., Huang, H., and Goldberg, K. Policy-based bayesian experimental design for non-differentiable implicit models.arXiv preprint arXiv:2203.04272,

    arXiv

  55. [55]

    Lindley, D. V . On a measure of the information provided by an experiment.The Annals of Mathematical Statistics, pp. 986–1005, 1956. 1, 2, 20

    1956

  56. [56]

    V .Bayesian statistics, a review, volume 2

    Lindley, D. V .Bayesian statistics, a review, volume 2. SIAM, 1972. 1, 2

    1972

  57. [57]

    Z., Haghgoo, B., Chen, A

    Liu, E. Z., Haghgoo, B., Chen, A. S., Raghunathan, A., Koh, P. W., Sagawa, S., Liang, P., and Finn, C. Just train twice: Improving group robustness without training group information. In Meila, M. and Zhang, T. (eds.),Proceedings of the 38th International Conference on Machine Learning, volume 139 ofProceedings of Machine Learning Research, pp. 6781–6792....

    2021

  58. [58]

    Deep learning face attributes in the wild.2015 IEEE International Conference on Computer Vision (ICCV), pp

    Liu, Z., Luo, P., Wang, X., and Tang, X. Deep learning face attributes in the wild.2015 IEEE International Conference on Computer Vision (ICCV), pp. 3730–3738, 2014. URL https://api.semanticscholar.org/CorpusID:459456. 8, 29

    2015

  59. [59]

    On the fairness of disentangled representations

    Locatello, F., Abbati, G., Rainforth, T., Bauer, S., Scholkopf, B., and Bachem, O. On the fairness of disentangled representations. InNeural Information Processing Systems, 2019. URL https://api.semanticscholar.org/CorpusID:173187920. 6

    2019

  60. [60]

    MacKay, D. J. Information-based objective functions for active data selection.Neural computa- tion, 4(4):590–604, 1992. 2, 6

    1992

  61. [61]

    Minimax pareto fairness: A multi objective perspective

    Martinez, N., Bertran, M., and Sapiro, G. Minimax pareto fairness: A multi objective perspective. InInternational conference on machine learning, pp. 6755–6764. PMLR, 2020. 7

    2020

  62. [62]

    Algorithmic fairness: Choices, assumptions, and definitions.Annual Review of Statistics and Its Application, 2021

    Mitchell, S., Potash, E., Barocas, S., D’Amour, A., and Lum, K. Algorithmic fairness: Choices, assumptions, and definitions.Annual Review of Statistics and Its Application, 2021. URL https://api.semanticscholar.org/CorpusID:228893833. 3, 6

    2021

  63. [63]

    I., Cavagnaro, D

    Myung, J. I., Cavagnaro, D. R., and Pitt, M. A. A tutorial on adaptive design optimization. Journal of mathematical psychology, 57(3-4):53–67, 2013. 20 13

    2013

  64. [64]

    Fairness without harm: An influence- guided active sampling approach.Advances in Neural Information Processing Systems, 37: 61513–61548, 2024

    Pang, J., Wang, J., Zhu, Z., Yao, Y ., Qian, C., and Liu, Y . Fairness without harm: An influence- guided active sampling approach.Advances in Neural Information Processing Systems, 37: 61513–61548, 2024. 7

    2024

  65. [65]

    Pytorch: An imperative style, high-performance deep learning library

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., Desmaison, A., Kopf, A., Yang, E., DeVito, Z., Raison, M., Tejani, A., Chilamkurthy, S., Steiner, B., Fang, L., Bai, J., and Chintala, S. Pytorch: An imperative style, high-performance deep learning library. InAdvances in Neural Informa...

    2019

  66. [66]

    On nesting monte carlo estimators

    Rainforth, T., Cornish, R., Yang, H., Warrington, A., and Wood, F. On nesting monte carlo estimators. InInternational Conference on Machine Learning, pp. 4267–4276. PMLR, 2018. 2, 20

    2018

  67. [67]

    R., and Bickford Smith, F

    Rainforth, T., Foster, A., Ivanova, D. R., and Bickford Smith, F. Modern bayesian experimental design.Statistical Science, 39(1):100–114, 2024. 1, 2, 20

    2024

  68. [68]

    M., Machado, J., and Baptista, L

    Realinho, V ., Vieira, M. M., Machado, J., and Baptista, L. Predict Students’ Dropout and Academic Success. UCI Machine Learning Repository, 2021. DOI: https://doi.org/10.24432/C5MC89. 8

    work page doi:10.24432/c5mc89 2021

  69. [69]

    C., and Fei-Fei, L

    Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., Berg, A. C., and Fei-Fei, L. Imagenet large scale visual recognition challenge.International Journal of Computer Vision (IJCV), 115(3):211–252, 2015. 29

    2015

  70. [70]

    G., Drovandi, C

    Ryan, E. G., Drovandi, C. C., Thompson, M. H., and Pettitt, A. N. Towards bayesian experimental design for nonlinear models that require a large number of sampling times. Computational Statistics & Data Analysis, 70:45–60, 2014. ISSN 0167-9473. doi: https: //doi.org/10.1016/j.csda.2013.08.017. 20

    work page doi:10.1016/j.csda.2013.08.017 2014

  71. [71]

    G., Drovandi, C

    Ryan, E. G., Drovandi, C. C., McGree, J. M., and Pettitt, A. N. A review of modern computa- tional algorithms for bayesian optimal design.International Statistical Review, 84(1):128–154,

  72. [72]

    Active learning literature survey

    Settles, B. Active learning literature survey. Technical report, University of Wisconsin— Madison, 2009. 6

    2009

  73. [73]

    Promoting fairness in learned models by learning to active learn under parity constraints

    Sharaf, A., Daume III, H., and Ni, R. Promoting fairness in learned models by learning to active learn under parity constraints. InProceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, pp. 2149–2156, 2022. 1, 7

    2022

  74. [74]

    Adaptive sampling for minimax fair classification.Advances in Neural Information Processing Systems, 34:24535–24544, 2021

    Shekhar, S., Fields, G., Ghavamzadeh, M., and Javidi, T. Adaptive sampling for minimax fair classification.Advances in Neural Information Processing Systems, 34:24535–24544, 2021. 6

    2021

  75. [75]

    Metric-fair active learning

    Shen, J., Cui, N., and Wang, J. Metric-fair active learning. InInternational conference on machine learning, pp. 19809–19826. PMLR, 2022. 1, 6, 7

    2022

  76. [76]

    Discrimination in online ad delivery.Commun

    Sweeney, L. Discrimination in online ad delivery.Commun. ACM, 56(5):44–54, May 2013. ISSN 0001-0782. doi: 10.1145/2447976.2447990. URL https://doi.org/10.1145/ 2447976.2447990. 1

    work page doi:10.1145/2447976.2447990 2013

  77. [77]

    H., Zhang, H., Park, J., Rong, K., and Whang, S

    Tae, K. H., Zhang, H., Park, J., Rong, K., and Whang, S. E. Falcon: Fair active learning using multi-armed bandits, 2024. URLhttps://arxiv.org/abs/2401.12722. 7

    arXiv 2024

  78. [78]

    Vincent, B. T. and Rainforth, T. The DARC toolbox: automated, flexible, and efficient delayed and risky choice experiments using bayesian adaptive design.PsyArXiv preprint, 2017. 20

    2017

  79. [79]

    Williams, D. R. . M. S. A. Discrimination and racial disparities in health: evidence and needed research.Journal of behavioral medicine, 32(1), 20–47., 2009. URL https://doi.org/10. 1007/s10865-008-9185-0. 1 14

    2009

  80. [80]

    2405.20674,https://arxiv.org/abs/2405.20674

    Xia, Y ., Mukherjee, S., Xie, Z., Wu, J., Li, X., Aponte, R., Lyu, H., Barrow, J., Chen, H., Dernoncourt, F., Kveton, B., Yu, T., Zhang, R., Gu, J., Ahmed, N. K., Wang, Y ., Chen, X., Deilamsalehy, H., Kim, S., Hu, Z., Zhao, Y ., Lipka, N., Yoon, S., Huang, T. K., Wang, Z., Mathur, P., Pal, S., Mukherjee, K., Zhang, Z., Park, N., Nguyen, T. H., Luo, J., R...

    work page internal anchor Pith review doi:10.48550/arxiv 2025

Showing first 80 references.