Pith. sign in

REVIEW 3 major objections 5 minor 41 references

Even modest noise on the validation set compresses training-data Shapley values toward zero by reshuffling which training points count as neighbors.

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 ↗

In-distribution validation noise directionally compresses KNN-Shapley values of training samples toward zero via neighborhood reshuffling, and a boundary-aware rescaling can partially restore baseline statistics.

T0 review reviewed 2026-07-12 challenge →

load-bearing objection Real, reproducible KNN-Shapley compression under validation noise; the general-stability claim is still an extrapolation. the 3 major comments →

arxiv 2607.03675 v1 pith:Y6R7ALTH submitted 2026-07-04 cs.LG

Validation-Induced Shapley Shifts: How Validation Structure Distorts Data Valuation

Yinan Shen , Ziao Yang , Hongfu Liu This is my paper
classification cs.LG
keywords data valuationShapley valueKNN-Shapleyvalidation setneighborhood reshufflingboundary samplesdata marketplacesnoise robustness
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.

The reading

Shapley values for training data are usually treated as fixed once the training set and model are chosen. This paper shows that assumption fails under ordinary, in-distribution changes to the validation set. Adding feature noise to validation points causes the Shapley values of training points to shrink consistently toward zero. The mechanism is geometric: noise reorders the local neighbors between each validation point and the training set, which flattens the marginal-contribution landscape. Using the efficient KNN-Shapley setting, the authors demonstrate the same directional compression on synthetic Gaussians and six real datasets. They then separate validation points into boundary and non-boundary groups by neighbor-label entropy and apply a simple rescaling that restores spread and positive-count statistics. The result is a practical warning for anyone pricing or selecting data with Shapley values: the numbers are only as stable as the validation geometry that defines them.

Core claim

Even modest in-distribution structural changes to the validation set, especially feature noise, produce consistent directional shifts in KNN-Shapley values of training samples, systematically compressing those values toward zero. The driver is a noise-induced neighborhood reshuffling effect that alters the local rank order between validation and training samples and thereby flattens the valuation landscape.

What carries the argument

Noise-induced neighborhood reshuffling, observed inside Soft-label KNN-Shapley: perturbations change which training points are the K nearest neighbors of each validation point, which in turn changes every training point's average marginal contribution. Boundary versus non-boundary validation groups (defined by neighbor-label entropy) separate the two main statistical effects—lost spread and lost positive count—so each can be rescaled separately.

Load-bearing premise

The paper treats the KNN-Shapley geometry with a fixed small neighborhood size and an entropy-based boundary mask as representative enough of Shapley-based data valuation in general that the observed compression and the proposed correction carry over beyond this testbed.

What would settle it

Run the same noise-injection protocol with a non-KNN utility (for example exact model-retraining Shapley on a small neural net or logistic regression) and check whether the same directional compression of training-point values and the same boundary/non-boundary split of effects still appear; if they do not, the claimed generality fails.

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

Share X Bluesky LinkedIn Reddit HN

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

3 major / 5 minor

Summary. The paper claims that Shapley values for training data are not stable under fixed training data and model once the validation set is altered by modest in-distribution structural changes (especially additive feature noise). Using Soft-label KNN-Shapley (K=5), the authors show on a 2-D Gaussian-quantiles synthetic set and six real datasets that noise systematically compresses training-sample Shapley values toward zero, reduces their standard deviation, and lowers the count of positive values. They attribute the effect to a noise-induced neighborhood reshuffling that alters local validation-to-training rank order, with distinct roles for boundary versus non-boundary validation points (isolated via one- and two-point case studies). They propose a boundary/non-boundary (B/N) group-wise rescaling plus positive-ratio bias correction (Appendix B) that restores baseline spread and sign structure when a clean reference is available, and discuss entropy-aware normalization and boundary-aware validation design for marketplace robustness.

Significance. If the directional compression and its geometric mechanism hold beyond the KNN setting, the result identifies a previously under-appreciated axis of fragility in data valuation that is directly relevant to collaborative learning, federated settings, and data marketplaces where validation protocols are decentralized or noisy. The empirical consistency across synthetic and real data, the fine-grained boundary/non-boundary case analysis, and the explicit geometric story (neighbor reordering flattening marginal utilities) are genuine strengths. The work does not claim a new algorithm but a diagnostic phenomenon; that framing is useful provided the scope is accurately delimited. The B/N correction, while partly by construction, supplies a concrete, reproducible mitigation template that future work can stress-test.

major comments (3)
  1. [Abstract, §1, §4–§5, §7] Abstract, §1, and §7 frame the result as a general challenge to “the assumption of Shapley stability” and as revealing “a new axis of fragility in data valuation.” All experiments, mechanism analysis (Figs. 1–5), and the B/N correction (Appendix B, Eqs. 6–11) are obtained exclusively with Soft-label KNN-Shapley (K=5). For KNN the utility of a validation point is literally a function of the ordered neighbor list, so any perturbation that reorders that list necessarily changes marginal contributions; the observed compression is therefore almost tautological for this utility class. Without at least one non-neighbor utility (logistic regression, small MLP, or fixed-kernel smoother) the marketplace-facing generality claim over-reaches the evidence. Either restrict the claims to KNN-Shapley or add the missing experiments.
  2. [§6, Appendix B] The proposed correction (Eqs. 3–5 and Appendix B Eqs. 6–11) defines α_B, α_N as ratios of clean-baseline to noisy group standard deviations and chooses the bias b by quantile matching so that the positive fraction exactly recovers the clean baseline ρ_bl. Consequently the restored std and positive-count statistics are recovered largely by construction once a clean reference distribution is known. In the very settings the paper worries about (noisy or adversarially designed validation), that clean baseline is unavailable. The practical utility of the method therefore remains unclear; the manuscript should either supply a baseline-free alternative or explicitly state the strong assumption.
  3. [Appendix B, §5] Boundary membership is defined by an entropy-of-neighbor-labels rule on the K=5 nearest training neighbors (Appendix B), yet no numerical threshold, sensitivity analysis, or ablation on the entropy cutoff appears. Because the entire B/N decomposition and the subsequent group-wise rescaling rest on this binary mask, the load-bearing geometric story is only partially specified. A precise definition and a short sensitivity check are required for reproducibility.
minor comments (5)
  1. [§4, Fig. 1] Figure 1 caption and surrounding text speak of “the cost of data protection,” but the experimental noise is isotropic Gaussian feature noise; the link to privacy mechanisms is never made precise.
  2. [Abstract, §7] The abstract and conclusion mention “entropy-aware normalization” as a mitigation, yet the body only develops the B/N rescaling; either expand the former or remove the claim.
  3. [Appendix A, §6] Table 1 and Appendix A list six real datasets; it would help readers if the main text briefly stated train/val sizes and the exact noise schedule (σ values) used for each.
  4. [§1, §5] Typographical inconsistencies appear (“neighborhood reshuffling as the keymechanism”, “We offeractionsfor fair”, missing spaces after periods in several places).
  5. [§2] References [13] and [11] are central to the KNN-Shapley lineage; a short comparison of how the present inflation/compression phenomenon relates to the inflation bias studied in [13] would strengthen the related-work discussion.

Circularity Check

1 steps flagged

Main shift phenomenon is empirical and non-circular; B/N correction recovers std and positive-count by explicit construction via α and quantile-matched b

specific steps
  1. fitted input called prediction [Section 6 Eqs. (3)–(5) and Appendix B Eqs. (8)–(11), Figures 6–8]
    "The group-wise scaling factors are defined by α_G = σ_bl_G / σ_noisy_G … We set b=−Q_{1−ρ_bl}({¯s_i}), ˜s_i = ¯s_i + b. This nearest-quantile matching ensures that the corrected scores recover the baseline fraction of positive Shapley values, while the group-wise scaling restores the spread … Figures 6–8 show that this correction substantially reduces the gap … tracks the baseline much more closely in both standard deviation and positive-count statistics, indicating that the boundary/non-boundary decomposition captures the main direction of the validation-induced shift."

    α_B, α_N and b are constructed precisely so that the corrected group spreads equal the baseline spreads and the positive ratio equals ρ_bl. Reporting recovery of exactly those two statistics is therefore tautological; the figures and the “indicating” claim restate the fitting objectives rather than independently validating the correction.

full rationale

The core claim (noise induces directional compression of KNN-Shapley values via neighborhood reshuffling) is an empirical observation demonstrated on synthetic Gaussian quantiles and six real datasets under Soft-label KNN-Shapley (K=5). It does not reduce by definition to its inputs; the mechanism is diagnosed from case studies of boundary vs non-boundary validation points and is falsifiable. The only circularity appears in the proposed mitigation (Section 6 + Appendix B). There α_B and α_N are defined exactly as the ratios of baseline to noisy group standard deviations, and the additive bias b is chosen by nearest-quantile matching so that the positive fraction of the corrected scores equals the clean baseline ρ_bl. Consequently, the paper’s claim that the correction “tracks the baseline much more closely in both standard deviation and positive-count statistics” is true by construction for those two metrics and does not constitute independent evidence that the B/N decomposition captures the shift. No load-bearing self-citation uniqueness theorem or ansatz smuggling is present; the minor self-citation to the authors’ prior inflation paper is only contextual. Overall circularity is therefore partial and confined to the evaluation of the corrective action.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 3 invented entities

The load-bearing content is empirical and geometric rather than axiomatic derivation. The claim rests on the standard Shapley/KNN utility definitions, a hand-chosen neighbor-entropy boundary rule, fixed K, and correction parameters explicitly fitted to restore clean-baseline moments. The main invented conceptual entities are the named phenomenon and the neighborhood-reshuffling mechanism; they are supported only by the paper's own experiments.

free parameters (5)
  • K (nearest neighbors) = 5
    Fixed to K=5 for all experiments and for the entropy boundary mask; changes K would change both valuations and the B/N partition.
  • noise level σ
    Gaussian validation noise scale is varied by the authors to induce the reported shifts; the phenomenon is demonstrated across chosen σ levels rather than derived.
  • group scaling factors α_B, α_N = σ_bl_G / σ_noisy_G
    Defined as σ_bl / σ_noisy for boundary and non-boundary groups so noisy spread matches baseline by construction (Appendix B).
  • bias b (positive-ratio quantile match) = -Q_{1-ρ_bl}({s̄_i})
    Chosen as the negative of the (1-ρ_bl)-quantile of corrected scores so the fraction of positive values equals the clean baseline.
  • boundary fraction λ = n_noisy_B / n_val
    Set to the noisy validation set's boundary fraction and used as the mixture weight between corrected B and N group scores.
axioms (4)
  • standard math Shapley value of a training point equals its expected marginal utility over coalitions, with utility equal to average negative loss on the validation set.
    Section 3, Eqs. (1)–(2); classical cooperative-game definition imported into data valuation.
  • domain assumption For KNN models, exact or soft-label Shapley values can be computed from neighbor orderings and serve as a faithful testbed for data valuation behavior.
    Section 3 and related work; paper uses Soft-label KNN-SV as computational backbone for all claims about 'Shapley' fragility.
  • ad hoc to paper A validation sample is a boundary sample iff the label entropy of its K nearest training neighbors is high enough to set the binary mask m_j=1.
    Appendix B introduces this entropy rule to partition columns for the B/N correction; it is a design choice of the paper, not a standard theorem.
  • domain assumption In-distribution additive Gaussian feature noise on validation inputs is a representative structural perturbation for studying validation-induced valuation shifts.
    Section 4 operationalizes 'modest changes to the validation set' almost entirely as ϵ~N(0,σ²I) on validation features.
invented entities (3)
  • validation-induced Shapley shifts no independent evidence
    purpose: Name the claimed systematic, directional change in training Shapley distributions caused by validation structure changes.
    Central named phenomenon of the paper; evidence is the authors' own synthetic and real-data experiments.
  • noise-induced neighborhood reshuffling effect no independent evidence
    purpose: Provide the geometric mechanism linking validation perturbations to flattened marginal utilities via reordered local neighbor ranks.
    Introduced in abstract/intro and used to unify boundary vs non-boundary case studies; not independently measured outside this manuscript.
  • boundary-aware B/N correction (group rescaling + positive-ratio bias) no independent evidence
    purpose: Mitigate noisy-validation distortion by separately restoring boundary spread and non-boundary positive mass toward a clean baseline.
    Section 6 and Appendix B define the procedure; success metrics are largely the same statistics used to set α and b.

reviewed 2026-07-12 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Validation-Induced Shapley Shifts: How Validation Structure Distorts Data Valuation." pith.science (2026). https://pith.science/paper/Y6R7ALTH

@misc{pith2026260703675,
  author       = {Pith},
  title        = {Pith review of: Validation-Induced Shapley Shifts: How Validation Structure Distorts Data Valuation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y6R7ALTH}},
  note         = {Machine review of arXiv:2607.03675}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Shapley values are widely used to attribute value to training data based on their marginal contribution to performance on a validation set. Existing practice often assumes these values are stable once the training data and model are fixed. In this work, we uncover a systematic vulnerability: even modest changes to the validation set, such as introducing noises, cause directional shifts in Shapley distributions. As noises are added, Shapley values of training samples compress toward zero. We trace this to a noise-induced neighborhood reshuffling effect: perturbations alter the local rank order between validation and training samples, flattening the valuation landscape. Using the KNN-Shapley framework, we show through synthetic and real data that these shifts are consistent and reproducible. Our findings challenge the assumption of Shapley stability and reveal a new axis of fragility in data valuation. We propose normalization and boundary-aware validation strategies to mitigate these distortions and enable more robust, interpretable valuation in machine learning marketplaces.

Figures

Figures reproduced from arXiv: 2607.03675 by Hongfu Liu, Yinan Shen, Ziao Yang.

Figure 1
Figure 1. Figure 1: Noise injection demo. A validation sample near the decision boundary is perturbed by Gaussian noise. Colors [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Shapley value shift under the noisy validation set. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Case I with one validation sample. Boundary sample vs non-boundary sample. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Case II with two validation samples [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Case II with two validation samples. the number of training points with net positive value. This mechanism provides a unified explanation for the observed distributional shifts and highlights the critical role of validation-set geometry in shaping data valuation. Our findings suggest that Shapley values are not purely intrinsic properties of training data, but are tightly coupled with the structure of the … view at source ↗
Figure 6
Figure 6. Figure 6: Shapley value correction on the Gaussian dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Shapley value correction on the real-world datasets: Phoneme, Planes2D, and Pol. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Shapley value correction on the real-world datasets: CPU, News20, and CreditCard. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗

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

41 extracted references · 10 linked inside Pith

  1. [1]

    ingredients

    Rachael Hwee Ling Sim, Xinyi Xu, and Bryan Kian Hsiang Low. Data valuation in machine learning:" ingredients", strategies, and open challenges. InIJCAI, pages 5607–5614, 2022

    2022

  2. [2]

    Data shapley: Equitable valuation of data for machine learning

    Amirata Ghorbani and James Zou. Data shapley: Equitable valuation of data for machine learning. InInternational Conference on Machine Learning, 2019

    2019

  3. [3]

    Collaborative machine learning with incentive-aware model rewards

    Rachael Hwee Ling Sim, Yehong Zhang, Mun Choon Chan, and Bryan Kian Hsiang Low. Collaborative machine learning with incentive-aware model rewards. InInternational Conference on Machine Learning, 2020

    2020

  4. [4]

    A marketplace for data: An algorithmic solution

    Anish Agarwal, Munther Dahleh, and Tuhin Sarkar. A marketplace for data: An algorithmic solution. InACM Conference on Economics and Computation, 2019

    2019

  5. [5]

    Lloyd S. Shapley. A value for n-person games. In Harold W. Kuhn and Albert W. Tucker, editors,Contributions to the Theory of Games II, pages 307–317. Princeton University Press, 1953

    1953

  6. [6]

    Efficient task-specific data valuation for nearest neighbor algorithms.arXiv preprint arXiv:1908.08619, 2019

    Ruoxi Jia, David Dao, Boxin Wang, Frances Ann Hubis, Nezihe Merve Gurel, Bo Li, Ce Zhang, Costas J Spanos, and Dawn Song. Efficient task-specific data valuation for nearest neighbor algorithms.arXiv preprint arXiv:1908.08619, 2019

    Pith/arXiv arXiv 1908

  7. [7]

    Cs-shapley: class-wise shapley values for data valuation in classification.Advances in Neural Information Processing Systems, 35:34574–34585, 2022

    Stephanie Schoch, Haifeng Xu, and Yangfeng Ji. Cs-shapley: class-wise shapley values for data valuation in classification.Advances in Neural Information Processing Systems, 35:34574–34585, 2022

    2022

  8. [8]

    A principled approach to data valuation for federated learning

    Tianhao Wang, Johannes Rausch, Ce Zhang, Ruoxi Jia, and Dawn Song. A principled approach to data valuation for federated learning. InFederated Learning: Privacy and Incentive, pages 153–167. Springer, 2020

    2020

  9. [9]

    Data valuation for vertical federated learning: A model-free and privacy-preserving method.arXiv preprint arXiv:2112.08364, 2021

    Xiao Han, Leye Wang, Junjie Wu, and Xiao Fang. Data valuation for vertical federated learning: A model-free and privacy-preserving method.arXiv preprint arXiv:2112.08364, 2021

    Pith/arXiv arXiv 2021

  10. [10]

    Data shapley valuation for efficient batch active learning

    Amirata Ghorbani, James Zou, and Andre Esteva. Data shapley valuation for efficient batch active learning. In 2022 56th asilomar conference on signals, systems, and computers, pages 1456–1462. IEEE, 2022

    2022

  11. [11]

    efficient task-specific data valuation for nearest neighbor algorithms

    Jiachen T. Wang and Ruoxi Jia. A note on “efficient task-specific data valuation for nearest neighbor algorithms”. arXiv preprint arXiv:2304.04258, 2023

    Pith/arXiv arXiv 2023

  12. [12]

    Threshold knn-shapley: A linear- time and privacy-friendly approach to data valuation.Conference on Neural Information Processing Systems, 2023

    Jiachen T Wang, Yuqing Zhu, Yu-Xiang Wang, Ruoxi Jia, and Prateek Mittal. Threshold knn-shapley: A linear- time and privacy-friendly approach to data valuation.Conference on Neural Information Processing Systems, 2023

    2023

  13. [13]

    On the inflation of knn-shapley value.arXiv preprint arXiv:2405.17489, 2024

    Ziao Yang, Han Yue, Jian Chen, and Hongfu Liu. On the inflation of knn-shapley value.arXiv preprint arXiv:2405.17489, 2024

    Pith/arXiv arXiv 2024

  14. [14]

    Optimizing data shapley interaction calculation from o (2ˆ n) to o (tnˆ 2) for knn models.arXiv preprint arXiv:2304.01224, 2023

    Mohamed Karim Belaid, Dorra El Mekki, Maximilian Rabus, and Eyke Hüllermeier. Optimizing data shapley interaction calculation from o (2ˆ n) to o (tnˆ 2) for knn models.arXiv preprint arXiv:2304.01224, 2023

    Pith/arXiv arXiv 2023

  15. [15]

    Explanations for data repair through shapley values

    Daniel Deutch, Nave Frost, Amir Gilad, and Oren Sheffer. Explanations for data repair through shapley values. In ACM International Conference on Information and Knowledge Management, 2021

    2021

  16. [16]

    A distributional framework for data valuation

    Amirata Ghorbani, Michael Kim, and James Zou. A distributional framework for data valuation. InInternational Conference on Machine Learning, 2020

    2020

  17. [17]

    Towards efficient data valuation based on the shapley value

    Ruoxi Jia, David Dao, Boxin Wang, Frances Ann Hubis, Nick Hynes, Nezihe Merve Gürel, Bo Li, Ce Zhang, Dawn Song, and Costas J Spanos. Towards efficient data valuation based on the shapley value. InInternational Conference on Artificial Intelligence and Statistics, 2019

    2019

  18. [18]

    Shapley-value data valuation for semi-supervised learning

    Christie Courtnage and Evgueni Smirnov. Shapley-value data valuation for semi-supervised learning. In International Conference on Discovery Science, pages 94–108. Springer, 2021

    2021

  19. [19]

    Beta shapley: a unified and noise-reduced data valuation framework for machine learning.arXiv preprint arXiv:2110.14049, 2021

    Yongchan Kwon and James Zou. Beta shapley: a unified and noise-reduced data valuation framework for machine learning.arXiv preprint arXiv:2110.14049, 2021

    Pith/arXiv arXiv 2021

  20. [20]

    A unified approach to interpreting model predictions.Advances in Neural Information Processing Systems, 2017

    Scott M Lundberg and Su-In Lee. A unified approach to interpreting model predictions.Advances in Neural Information Processing Systems, 2017

    2017

  21. [21]

    Improving kernelshap: Practical shapley value estimation via linear regression.arXiv preprint arXiv:2012.01536, 2020

    Ian Covert and Su-In Lee. Improving kernelshap: Practical shapley value estimation via linear regression.arXiv preprint arXiv:2012.01536, 2020

    Pith/arXiv arXiv 2012

  22. [22]

    Trustworthy machine learning for health care: scalable data valuation with the shapley value

    Konstantin D Pandl, Fabian Feiland, Scott Thiebes, and Ali Sunyaev. Trustworthy machine learning for health care: scalable data valuation with the shapley value. InConference on Health, Inference, and Learning, 2021

    2021

  23. [23]

    Data valuation for medical imaging using shapley value and application to a large-scale chest x-ray dataset

    Siyi Tang, Amirata Ghorbani, Rikiya Yamashita, Sameer Rehman, Jared A Dunnmon, James Zou, and Daniel L Rubin. Data valuation for medical imaging using shapley value and application to a large-scale chest x-ray dataset. Scientific Reports, 11(1):8366, 2021. 11 Validation-Induced Shapley Shifts

    2021

  24. [24]

    Data banzhaf: A robust data valuation framework for machine learning

    Jiachen T Wang and Ruoxi Jia. Data banzhaf: A robust data valuation framework for machine learning. In International Conference on Artificial Intelligence and Statistics, 2023

    2023

  25. [25]

    Improving cooperative game theory-based data valuation via data utility learning.arXiv preprint arXiv:2107.06336, 2021

    Tianhao Wang, Yu Yang, and Ruoxi Jia. Improving cooperative game theory-based data valuation via data utility learning.arXiv preprint arXiv:2107.06336, 2021

    Pith/arXiv arXiv 2021

  26. [26]

    Efficient sampling approaches to shapley value approximation.ACM on Management of Data, 1(1):1–24, 2023

    Jiayao Zhang, Qiheng Sun, Jinfei Liu, Li Xiong, Jian Pei, and Kui Ren. Efficient sampling approaches to shapley value approximation.ACM on Management of Data, 1(1):1–24, 2023

    2023

  27. [27]

    Polynomial calculation of the shapley value based on sampling

    Javier Castro, Daniel Gómez, and Juan Tejada. Polynomial calculation of the shapley value based on sampling. Computers & Operations Research, 36(5):1726–1730, 2009

    2009

  28. [28]

    On shapley value in data assemblage under independent utility

    Xuan Luo, Jian Pei, Zicun Cong, and Cheng Xu. On shapley value in data assemblage under independent utility. arXiv preprint arXiv:2208.01163, 2022

    Pith/arXiv arXiv 2022

  29. [29]

    Efficient data shapley for weighted nearest neighbor algorithms

    Jiachen T Wang, Prateek Mittal, and Ruoxi Jia. Efficient data shapley for weighted nearest neighbor algorithms. InInternational Conference on Artificial Intelligence and Statistics, 2024

    2024

  30. [30]

    Do imagenet classifiers generalize to imagenet? InInternational Conference on Machine Learning, 2019

    Benjamin Recht, Rebecca Roelofs, Ludwig Schmidt, and Vaishaal Shankar. Do imagenet classifiers generalize to imagenet? InInternational Conference on Machine Learning, 2019

    2019

  31. [31]

    Collaborative machine learning markets with data-replication-robust payments.arXiv preprint arXiv:1911.09052, 2019

    Olga Ohrimenko, Shruti Tople, and Sebastian Tschiatschek. Collaborative machine learning markets with data-replication-robust payments.arXiv preprint arXiv:1911.09052, 2019

    Pith/arXiv arXiv 1911

  32. [32]

    The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients.Expert Systems with Applications, 36(2):2473–2480, 2009

    I-Cheng Yeh and Che-hui Lien. The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients.Expert Systems with Applications, 36(2):2473–2480, 2009. doi: 10.1016/j.eswa.2007.12.020

    doi:10.1016/j.eswa.2007.12.020 2009

  33. [33]

    Openml dataset: phoneme (version 1), 2014

    OpenML. Openml dataset: phoneme (version 1), 2014. URL https://www.openml.org/d/1489. Binary nasal-vs-oral phoneme classification dataset. OpenML metadata lists KEEL and ELENA as source repositories. Accessed 2026-04-12

    2014

  34. [34]

    Friedman, Richard A

    Leo Breiman, Jerome H. Friedman, Richard A. Olshen, and Charles J. Stone.Classification and Regression Trees. Wadsworth International Group, 1984

    1984

  35. [35]

    Openml dataset: 2dplanes (version 2), 2014

    OpenML. Openml dataset: 2dplanes (version 2), 2014. URL https://www.openml.org/d/727. Binarized OpenML version used in code. Accessed 2026-04-12

    2014

  36. [36]

    Openml dataset: pol (version 2), 2014

    OpenML. Openml dataset: pol (version 2), 2014. URL https://www.openml.org/d/722. Binarized OpenML version used in code. Accessed 2026-04-12

    2014

  37. [37]

    Delve data repository: Computer activity dataset, 1990

    DELVE Project. Delve data repository: Computer activity dataset, 1990. URLhttps://www.cs.toronto.edu/ ~delve/data/comp-activ/desc.html. Official dataset repository page for the Computer Activity dataset family. Accessed 2026-04-12

    1990

  38. [38]

    Openml dataset: cpu_act (version 3), 2014

    OpenML. Openml dataset: cpu_act (version 3), 2014. URL https://www.openml.org/d/761. Binarized OpenML version used in code. Accessed 2026-04-12

    2014

  39. [39]

    Newsweeder: Learning to filter netnews

    Ken Lang. Newsweeder: Learning to filter netnews. InMachine Learning Proceedings 1995, pages 331–339,

    1995

  40. [40]

    doi: 10.1016/B978-1-55860-377-6.50048-7

    doi:10.1016/b978-1-55860-377-6.50048-7

  41. [41]

    Sentence-bert: Sentence embeddings using siamese bert-networks

    Nils Reimers and Iryna Gurevych. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th Inter- national Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3982–3992, 2019. doi: 10.18653/v1/D19-1410. 12

    doi:10.18653/v1/d19-1410 2019

This paper was first reviewed by grok-4.5 on July 12, 2026.