Pith. sign in

REVIEW 4 major objections 5 minor 63 references

From Machine Learning to Machine Unlearning: Complying with GDPR's Right to be Forgotten while Maintaining Business Value of Predictive Models

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read ETID shows that machine unlearning can be fast, accurate, and consistent with naive retraining, all at once.

desk verdict A practical ensemble-plus-distillation unlearning framework with solid empirical results on two datasets; the theoretical wrapper overclaims consistency and the repeated-request case is unproven. read the letter →

arxiv 2411.17126 v2 pith:ZO6MBBVF submitted 2024-11-26 cs.LG

Yuncong Yang , Xiao Han , Yidong Chai , Reza Ebrahimi , Rouzbeh Behnia , Balaji Padmanabhan This is my paper · ORCID
classification cs.LG
keywords machineunlearningrighttobeforgottenGDPRcomplianceensemblelearningknowledgedistillationdataerasurepredictiveanalyticsmodelconsistency
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to show that honoring a data-erasure request need not mean paying for a full retraining or accepting a degraded predictive model. It proposes ETID, a two-stage framework: build an ensemble whose sub-models are trained on overlapping slices of the data, then erase any requested samples by distilling the affected sub-models toward sibling sub-models that have never seen those samples. The claim is that this achieves all four desiderata of machine unlearning—consistency with naive retraining, accuracy, efficiency, and verifiability—simultaneously, which previous methods trade off against one another. If correct, companies subject to GDPR's Right to Be Forgotten could erase data in seconds rather than retraining from scratch, while keeping service quality and avoiding fines for both non-compliance and business loss.

What carries the argument

The load-bearing object is the retrained-alike model: a model trained by the same algorithm on data that overlaps a target model's training set by at least a factor of one in shared versus unique samples, so that it can serve as a distillation reference for what the target would look like if retrained without a specific block of data. ROEL manufactures these references for free by training $K$ sub-models on leave-one-part-out subsets; TID then applies the standard distillation objective—minimizing the divergence between the reference's outputs and the target sub-model's outputs on the samples to be erased—iterates over all affected parts, rectifies the sub-models on the remaining data, and updates the references. The same overlap condition, stated as Proposition 2, governs whether the references remain valid after many rounds of unlearning.

What would settle it

Take a ROEL ensemble with K=5, erase one entire data part, and compare the TID-unlearned model against a model naively retrained without that part; if their L2 prediction distance is no smaller than the distance between two independent retrains on identical data, the retrained-alike premise is not doing the work claimed.

Watch

Extended reading notes

Core claim

The paper's central claim is that a carefully constructed ensemble makes unlearning nearly free without sacrificing the model's business value. With Reference-Oriented Ensemble Learning (ROEL), the training data is split into $K$ equal parts and $K$ sub-models are trained, each on all parts except one; because every pair of sub-models shares $(K-2)/K$ of its training data, the sub-model that excludes a requested sample's part is, by the paper's definitions, a 'retrained-alike' model of any sub-model that saw that part. The iterative information distillation stage (TID) then uses that sibling sub-model as a reference teacher, fine-tuning each affected sub-model so its outputs on the erased samples match the reference, rectifying with the remaining data, and updating the reference for future requests. The paper argues these steps erase the target samples' influence while keeping predictions close to what a naïve retrained model would have produced, and reports experiments on consumer profiling and image classification where ETID meets all four machine-unlearning desiderata simultaneously.

Load-bearing premise

The guarantee rests on treating two models trained on almost the same data as interchangeable teachers: for deep networks, high data overlap does not by itself ensure that the two models make nearly the same predictions.

Editorial extensions

If this is right

  • Deletion requests can be answered at a tiny fraction of retraining cost: ETID erases 1% of the training data in 8.63 seconds on the Purchase task and 958 seconds on CIFAR-100 with parallel computing, versus 43.6 and 8,295 seconds for single-model naïve retraining.
  • The unlearned model stays accurate for remaining, test, and unlearning samples, with test accuracies of 0.952 on Purchase and 0.776 on CIFAR-100, exceeding the best comparison methods.
  • The distilled model's L2 prediction distance to a naïve retrained model is smaller than that of the leading non-retrained baselines on both datasets, so consistency and compliance are not bought at the cost of accuracy.
  • Membership-inference AUC drops significantly after unlearning, so the erased samples are no longer recognizable as training members, satisfying verifiability.
  • Because the ensemble's sub-models are built to be retrained-alike from the start, the same model stock serves future requests without extra reference training, as long as the overlap condition in Proposition 2 holds.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: each unlearning round that preserves the Proposition 2 overlap condition refreshes the reference models, so the framework amortizes across many deletion requests without a full retraining.
  • Editorial inference: a natural stress test is to report per-sample consistency on rare classes and on the erased samples themselves, since the paper's consistency numbers are aggregate L2 distances over all predictions.
  • Editorial inference: the same overlap-and-distill recipe could be ported to distributed model training by letting participants share a small common anchor set, so one participant's model can serve as another's erasure reference.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes a holistic machine learning-to-unlearning framework, ETID, consisting of a reference-oriented ensemble learning method (ROEL) for model construction and an iterative information distillation method (TID) for responding to data-erasure requests. ROEL trains K overlapping sub-models, each on a leave-one-part-out subset, and TID distills unlearning data from affected sub-models under the supervision of sub-models that are claimed to be 'retrained-alike' references, then rectifies models on remaining data. The paper claims that ETID simultaneously achieves consistency, accuracy, efficiency, and verifiability, and presents experiments on two datasets (Purchase and CIFAR100) comparing against several unlearning baselines.

Significance. If the claims are established, ETID would be a practically valuable method for GDPR-style right-to-be-forgotten compliance, potentially enabling efficient data erasure while preserving predictive accuracy. The paper is among the first to frame machine unlearning as a holistic two-stage business problem, and the empirical study covers multiple desiderata and includes parameter sensitivity analysis. The framework is simple and reproducible in principle. However, the theoretical guarantees are substantially weaker than the abstract suggests: the 'retrained-alike' notion is defined purely by data overlap, not behavioral similarity, and the proofs of Propositions 2 and 3 rely on unjustified assumptions. The empirical results on two datasets are encouraging but do not compensate for the missing theoretical support for the central consistency claim.

major comments (4)
  1. [§4.1, Definitions 3–4 and Proposition 1] The 'retrained-alike' notion is defined solely by the proportion of shared training samples, with no behavioral condition on the learning algorithm A. Proposition 1 therefore establishes only a data-overlap fact: Mi shares (K−2)/K of its data with the true retrained model M_rt^j. For deep networks (MLPs and ResNet18 in the experiments), data overlap does not imply that predictions are close on any input distribution, and no quantitative bound is given. Since the TID objective (Eq. 3) aligns outputs only on the unlearning data Xu, nothing in the construction constrains the unlearned model to match M_rt^j on remaining or test data. The abstract's claim that ETID delivers predictions 'consistent with those of naïve retrained models' is therefore not supported by the proof structure.
  2. [§4.2.4, Proposition 2] After a first unlearning request, Eq. 5 updates each reference model to the corresponding unlearned sub-model, which has been modified by distillation (Eq. 3) and rectification (Eq. 4). Such a model is no longer of the form A(Dr_-i), i.e., it is not a freshly retrained model on the remaining data. The proof of Proposition 2 nonetheless treats the updated reference as if it were freshly retrained, deriving an if-and-only-if condition (Eq. 6) purely on data overlap. The condition therefore does not characterize the object that actually serves as the reference in the next round, leaving the iterative consistency claim underived.
  3. [§4.3, Proposition 3] The proof of distinguishability assumes that Mj(Xu) ≠ Mi(Xu). This inequality is not guaranteed; in fact, if Mi is a good retrained-alike reference, its outputs on unlearning data may coincide with those of Mj. The argument that the difference between M^u_j(Xu) and Mj(Xu) can serve as a verification function therefore requires a behavioral separation that the paper does not establish. As a result, verifiability is not proven by the given reasoning.
  4. [§5.2.2, Table 7] The consistency results are aggregate L2 distances between the unlearned and true retrained models, computed on two datasets. These empirical numbers are favorable, but they cannot substitute for the missing theoretical guarantee, because the proofs in Section 4 only establish data-overlap relations, not output-level closeness. The paper should either provide a rigorous behavioral bound connecting data overlap to prediction proximity for the algorithm class used, or explicitly reframe the consistency claim as an empirical finding rather than a proven property.
minor comments (5)
  1. [§4.1] The section title contains a typo: 'Reference-Oriented Ensemble Leaning' should be 'Learning'. The abbreviation 'E.q.' appears in several equations (e.g., Eq. 3 and Eq. 4) and should be 'Eq.'.
  2. [Tables 6, 9, 11–14] Several table cells contain repeated values in a single cell (e.g., '0.999 ± 0.0010.999 ± 0.0010.999 ± 0.001' in Table 6 and '8.63 ± 0.248.63 ± 0.248.63 ± 0.24' in Table 9). These appear to be copy-paste formatting errors and should be corrected.
  3. [§5.1 and §5.2.6] The default K=5 is justified by a reference to Breiman (1996), but the sensitivity analysis does not connect the degradation at large unlearning ratios to the condition in Eq. 6. A discussion of when Proposition 2's condition fails (e.g., for UR=10% on either dataset) and how the recommended retraining-from-scratch remedy applies would make the practical guidance more concrete.
  4. [§5 (entire experimental section)] No code or detailed hyperparameters are provided (e.g., learning rates, number of epochs for distillation and rectification, optimizer choices). Given that the paper makes quantitative claims about efficiency and consistency, releasing an implementation or at least full training details would materially improve reproducibility.
  5. [§6] The statement that adopting ETID could help companies 'save up to EUR 20 million' is speculative and not derived from the experiments. The managerial implications would be stronger if such quantitative claims were either removed or accompanied by a clearly stated model and assumptions.

Circularity Check

2 steps flagged · score 4.0 of 10

The 'retrained-alike' property is stipulated by data-overlap definitions, and Proposition 2's proof substitutes the true retrained model for the updated unlearned reference; the empirical benchmark comparisons keep the headline consistency results independent.

  1. self definitional [Section 4.1, Definitions 3-4 and Proposition 1 (pp. 15-16)]
    "Definition 4 (Retrained-alike model) Given a model M = A(D), unlearning data Du ⊂ D, and the naïve retrained model M rt = A(D/Du), any δ-alike (δ ≥ 1) model of M rt that has not been trained on Du is a retrained-alike model of M. ... Proposition 1 Given the sub-models generated by ROEL and unlearning data Du ⊂ di, sub-model Mi is a retrained-alike model of sub-model Mj, ∀j ̸= i."

    Definition 3 defines δ-alike purely by the ratio of shared training samples, and Definition 4 then declares any δ-alike (δ≥1) model not trained on Du to be 'retrained-alike'. Proposition 1's proof only verifies that Mi and M^rt_j share D−ij and each has one unique part, so δ≥1; no output-level closeness to the naïve retrained model is shown or bounded. The proposition is therefore true by the authors' own stipulative definition, and the label 'retrained-alike' is attached by construction. Sections 4.2 and 4.3 then use this label as the basis for claiming that TID can 'ensure unlearning consistency', transferring a definitional data-overlap property onto the behavioral desideratum of Definition 1 without a derivational bridge.

  2. other [Section 4.2.4 (Eq. 5-6) and Appendix A.2 (proof of Proposition 2)]
    "At the end of the unlearning process, we update the reference models by replacing them with the corresponding unlearned sub-models: M rf_i = Mi, ∀i ∈ [1, K] & du_i ̸= ∅. ... Proposition 2 ... M rf_i can still be considered a retrained-alike model of sub-model Mj, ∀j ̸= i, if and only if: [Eq. 6]. ... Proof. ... We can deduce that updated reference model M rf_i is a δ-alike (δ ≥ 1) model of M rt_j (i.e., the naïve retrained version of Mj)."

    Eq. 5 defines the updated reference as the unlearned sub-model, which has been modified by distillation (Eq. 3) and rectification (Eq. 4), so it is generally not a model of the form A(Dr−i). The proposition's condition (Eq. 6) and its proof, however, count the training data of M^rt_i and Mj as if M^rf_i were freshly retrained on Dr−i. The proof simply asserts that M^rf_i is a δ-alike model of M^rt_j instead of deriving it from the actual construction; the iff condition therefore characterizes the true retrained model, not the updated reference. This assumes the very equivalence (updated reference = retrained model) that the unlearning method is supposed to approximate without retraining.

full rationale

The paper is not a case of fitted-input-called-prediction: the consistency metric Con(·) in Eq. 8 is computed against truly retrained naïve models, and the accuracy, efficiency, and verifiability results are external benchmark comparisons. Thus the headline empirical claims have independent content and are not forced by the reference construction. The circularity is confined to the formal scaffolding. First, Definitions 3-4 make 'retrained-alike' a data-overlap label that ROEL satisfies by construction, so Proposition 1 is a definitional tautology rather than an output-level guarantee; later sections lean on this label to justify consistency. Second, Proposition 2's proof silently replaces the updated reference defined by Eq. 5 with the freshly retrained model when deriving Eq. 6, so the if-and-only-if condition characterizes the object the method is trying to avoid computing, not the reference actually used. These are load-bearing proof steps for the framework's formal claims, but they do not generate the reported L2 consistency numbers, so the overall circularity is moderate rather than complete.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the availability of unlearning data during distillation, on the behavioral proxy that data-overlap implies output similarity, and on the standard unlearning convention that deletion requests concern training samples. No physical entities are introduced. The only hand-set hyperparameter is K, the number of sub-models.

free parameters (1)
  • K (number of ROEL sub-models) = 5 (default); 3, 7, 10 in sensitivity analysis
    K determines the overlap ratio (K-2 shared parts vs 1 unique part) used to claim retrained-alikeness in Proposition 1; it is chosen by the experimenter, not derived from data.
assumptions (5)
  • domain assumption Unlearning requests concern samples in the training set D
    Stated in footnote 7 and used throughout the problem formulation; follows prior unlearning literature but is not required by GDPR.
  • ad hoc to paper A model trained on D_-i, which shares (K-2)/K of its data with another model, is behaviorally close to that model's naive retrained version
    This is the retrained-alike assumption; Definition 4 makes it a data-overlap condition, but the proofs require output-level closeness.
  • domain assumption Distillation against reference outputs on Du removes the requested information
    Core premise of all distillation-based unlearning (Section 2.2.2); ETID inherits it without a new argument.
  • domain assumption Membership inference AUC difference is an adequate verifiability test
    Used in Section 5.2.5 following prior work (Ma et al. 2022, Xu et al. 2023).
  • standard math Equal-size partition and K>=3 yield delta>=1 in the alikeness bound
    Used in the proof of Proposition 1 in Appendix A.1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Machine Learning to Machine Unlearning: Complying with GDPR's Right to be Forgotten while Maintaining Business Value of Predictive Models." pith.science (2026). https://pith.science/paper/ZO6MBBVF

@misc{pith2026241117126,
  author       = {Pith},
  title        = {Pith review of: From Machine Learning to Machine Unlearning: Complying with GDPR's Right to be Forgotten while Maintaining Business Value of Predictive Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZO6MBBVF}},
  note         = {Machine review of arXiv:2411.17126}
}
read the original abstract

Recent privacy regulations (e.g., GDPR) grant data subjects the `Right to Be Forgotten' (RTBF) and mandate companies to fulfill data erasure requests from data subjects. However, companies encounter great challenges in complying with the RTBF regulations, particularly when asked to erase specific training data from their well-trained predictive models. While researchers have introduced machine unlearning methods aimed at fast data erasure, these approaches often overlook maintaining model performance (e.g., accuracy), which can lead to financial losses and non-compliance with RTBF obligations. This work develops a holistic machine learning-to-unlearning framework, called Ensemble-based iTerative Information Distillation (ETID), to achieve efficient data erasure while preserving the business value of predictive models. ETID incorporates a new ensemble learning method to build an accurate predictive model that can facilitate handling data erasure requests. ETID also introduces an innovative distillation-based unlearning method tailored to the constructed ensemble model to enable efficient and effective data erasure. Extensive experiments demonstrate that ETID outperforms various state-of-the-art methods and can deliver high-quality unlearned models with efficiency. We also highlight ETID's potential as a crucial tool for fostering a legitimate and thriving market for data and predictive services.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 57 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution isbn issn journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.all := #1...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in "" FUNCTION format.date year ...

  3. [3]

    Accenture. (2018). Making it personal: Why brands must move from communication to conversation for greater personalization. https://tinyurl.com/4hbd746m

    work page 2018

  4. [4]

    Acquisti, A., Brandimarte, L., & Loewenstein, G. (2015). Privacy and human behavior in the age of information. Science, 347(6221), 509-514

    work page 2015

  5. [5]

    Acquisti, A., Taylor, C., & Wagman, L. (2016). The economics of privacy. Journal of economic Literature, 54(2), 442-492

    work page 2016

  6. [6]

    Arpit, D., Jastrzębski, S., Ballas, N., Krueger, D., Bengio, E., Kanwal, M. S., ... & Lacoste-Julien, S. (2017, July). A closer look at memorization in deep networks. In International conference on machine learning (pp. 233-242). PMLR

    work page 2017

  7. [7]

    Auxier, B., Rainie, L., Anderson, M., Perrin, A., Kumar, M., & Turner, E. (2019). Americans and privacy: Concerned, confused and feeling lack of control over their personal information

    work page 2019

  8. [8]

    F., & Krishnan, M

    Awad, N. F., & Krishnan, M. S. (2006). The personalization privacy paradox: an empirical evaluation of information transparency and the willingness to be profiled online for personalization. MIS quarterly, 13-28

    work page 2006

Show all 63 references
  1. [9]

    Berente, N., Gu, B., Recker, J., & Santhanam, R. (2021). Managing artificial intelligence. MIS quarterly, 45(3)

    2021

  2. [10]

    A., Jia, H., Travers, A., Zhang, B.,

    Bourtoule, L., Chandrasekaran, V., Choquette-Choo, C. A., Jia, H., Travers, A., Zhang, B., ... & Papernot, N. (2021, May). Machine unlearning. In 2021 IEEE Symposium on Security and Privacy (SP) (pp. 141-159). IEEE

    2021

  3. [11]

    Breiman, L. (1996). Bagging predictors. Machine learning, 24, 123-140

    1996

  4. [12]

    (2015, May)

    Cao, Y., & Yang, J. (2015, May). Towards making systems forget with machine unlearning. In 2015 IEEE Symposium on Security and Privacy (SP) (pp. 463-480). IEEE

    2015

  5. [13]

    Chen, C., Sun, F., Zhang, M., & Ding, B. (2022a). Recommendation unlearning. In Proceedings of the ACM Web Conference 2022 (pp. 2768-2777)

    2022

  6. [14]

    Chen, M., Zhang, Z., Wang, T., Backes, M., Humbert, M., & Zhang, Y. (2022b). Graph unlearning. In Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security (pp. 499-513)

    2022

  7. [15]

    Chen, X., Simchi-Levi, D., & Wang, Y. (2022c). Privacy-preserving dynamic personalized pricing with demand learning. Management Science, 68(7), 4878-4898

    2022

  8. [16]

    S., Tarun, A

    Chundawat, V. S., Tarun, A. K., Mandal, M., & Kankanhalli, M. (2023, June). Can bad teaching induce forgetting? unlearning in deep networks using an incompetent teacher. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 37, No. 6, pp. 7210-7217)

    2023

  9. [17]

    O., & Kohli, R

    Cichy, P., Salge, T. O., & Kohli, R. (2021). Privacy Concerns and Data Sharing in the Internet of Things: Mixed Methods Evidence from Connected Cars. Mis Quarterly, 45(4)

    2021

  10. [18]

    (2006, July)

    Dwork, C. (2006, July). Differential privacy. In International colloquium on automata, languages, and programming (pp. 1-12). Berlin, Heidelberg: Springer Berlin Heidelberg

    2006

  11. [19]

    & Chen, J

    Gao, X., Ma, X., Wang, J., Sun, Y., Li, B., Ji, S., ... & Chen, J. (2024). Verifi: Towards verifiable federated unlearning. IEEE Transactions on Dependable and Secure Computing

    2024

  12. [20]

    Ghose, A., Li, B., Macha, M., Sun, C., & Foutz, N. Z. (2024). Privacy Choice during Crisis. Forthcoming in Management Science

    2024

  13. [21]

    Ginart, A., Guan, M., Valiant, G., & Zou, J. Y. (2019). Making ai forget you: Data deletion in machine learning. Advances in neural information processing systems, 32

    2019

  14. [22]

    Golatkar, A., Achille, A., & Soatto, S. (2020). Eternal sunshine of the spotless net: Selective forgetting in deep networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 9304-9312)

    2020

  15. [23]

    Golatkar, A., Achille, A., & Soatto, S. (2020). Forgetting outside the box: Scrubbing deep networks of information accessible from input-output observations. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXIX 16 (pp....

    2020

  16. [24]

    Golatkar, A., Achille, A., Ravichandran, A., Polito, M., & Soatto, S. (2021). Mixed-privacy forgetting in deep networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 792-801)

    2021

  17. [25]

    (2021, May)

    Graves, L., Nagisetty, V., & Ganesh, V. (2021, May). Amnesiac machine learning. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 35, No. 13, pp. 11516-11524)

    2021

  18. [26]

    & Van Der Maaten, L

    Guo, C., Goldstein, T., Hannun, A. & Van Der Maaten, L.. (2020). Certified Data Removal from Machine Learning Models. In International conference on machine learning (pp. 3832--3842). PMLR

    2020

  19. [27]

    Han, X., Wang, L., & Fan, W. (2021). Is hidden safe? Location protection against machine-learning prediction attacks in social networks. MIS quarterly, 45(2), 821-858

    2021

  20. [28]

    H., & Tsoukalas, G

    Hastings, M., Falk, B. H., & Tsoukalas, G. (2023). Privacy-preserving network analytics. Management Science, 69(9), 5482-5500

    2023

  21. [29]

    He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770-778)

    2016

  22. [30]

    R., March, S

    Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information systems research. MIS quarterly, 75-105

    2004

  23. [31]

    Hinton, G., Vinyals, O., & Dean, J. (2015). Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531

    2015 arXiv

  24. [32]

    A., Chaudhuri, K., & Zou, J

    Izzo, Z., Smart, M. A., Chaudhuri, K., & Zou, J. (2021, March). Approximate data deletion from machine learning models. In International Conference on Artificial Intelligence and Statistics (pp. 2008-2016). PMLR

    2021

  25. [33]

    A., Shriver, S

    Johnson, G. A., Shriver, S. K., & Goldberg, S. G. (2023). Privacy and market concentration: intended and unintended consequences of the GDPR. Management Science, 69(10), 5695-5721

    2023

  26. [34]

    T., & Sudhir, K

    Ke, T. T., & Sudhir, K. (2023). Privacy rights and data security: GDPR and personal data markets. Management Science, 69(8), 4389-4412

    2023

  27. [35]

    KPMG. (2021). Corporate data responsibility: Bridging the consumer trust gap. KPMG, Amstelveen, Netherlands

    2021

  28. [36]

    Kurmanji, M., Triantafillou, P., Hayes, J., & Triantafillou, E. (2023). Towards unbounded machine unlearning. Advances in neural information processing systems, 36

    2023

  29. [37]

    Lei, Y., Miao, S., & Momot, R. (2023). Privacy-preserving personalized revenue management. Management Science, 70(7), 4875-4892

    2023

  30. [38]

    J., Yu, Y., & Gupta, S

    Li, S., Schneider, M. J., Yu, Y., & Gupta, S. (2023). Reidentification risk in panel data: Protecting for k-anonymity. Information Systems Research, 34(3), 1066-1088

    2023

  31. [39]

    B., & Qin, J

    Li, X. B., & Qin, J. (2017). Anonymizing and sharing medical text records. Information Systems Research, 28(2), 332-352

    2017

  32. [40]

    B., & Sarkar, S

    Li, X. B., & Sarkar, S. (2013). Class-restricted clustering and microperturbation for data privacy. Management Science, 59(4), 796-812

    2013

  33. [41]

    B., & Sarkar, S

    Li, X. B., & Sarkar, S. (2014). Digression and value concatenation to enable privacy-preserving regression. MIS quarterly: management information systems, 38(3), 679

    2014

  34. [42]

    & Liu, Y

    Liu, S., Yao, Y., Jia, J., Casper, S., Baracaldo, N., Hase, P., ... & Liu, Y. (2024). Rethinking Machine Unlearning for Large Language Models. arXiv preprint arXiv:2402.08787

    2024 arXiv

  35. [43]

    Ma, Z., Liu, Y., Liu, X., Liu, J., Ma, J., & Ren, K. (2022). Learn to forget: Machine unlearning via neuron masking. IEEE Transactions on Dependable and Secure Computing, 20(4), 3194-3207

    2022

  36. [44]

    Z., Li, B., & Ghose, A

    Macha, M., Foutz, N. Z., Li, B., & Ghose, A. (2024). Personalized privacy preservation in consumer mobile trajectories. Information Systems Research, 35(1), 249-271

    2024

  37. [45]

    Menon, S., & Sarkar, S. (2016). Privacy and Big Data: Scalable Approaches to Sanitize Large Transactional Databases for Sharing. MIS Quarterly, 40(4), 963-982

    2016

  38. [46]

    Pavlou, P. A. (2011). State of the information privacy literature: Where are we now and where should we go?. MIS quarterly, 977-988

    2011

  39. [47]

    Shmueli, G., & Koppius, O. R. (2011). Predictive analytics in information systems research. MIS quarterly, 553-572

    2011

  40. [48]

    J., Milberg, S

    Smith, H. J., Milberg, S. J., & Burke, S. J. (1996). Information privacy: Measuring individuals' concerns about organizational practices. MIS quarterly, 167-196

    1996

  41. [49]

    J., Dinev, T., & Xu, H

    Smith, H. J., Dinev, T., & Xu, H. (2011). Information privacy research: an interdisciplinary review. MIS quarterly, 989-1015

    2011

  42. [50]

    Song, Y., Sahoo, N., & Ofek, E. (2019). When and how to diversify—a multicategory utility model for personalized content recommendation. Management Science, 65(8), 3737-3757

    2019

  43. [51]

    H., & Phang, C

    Sutanto, J., Palme, E., Tan, C. H., & Phang, C. W. (2013). Addressing the personalization-privacy paradox: An empirical assessment from a field experiment on smartphone users. MIS quarterly, 1141-1164

    2013

  44. [52]

    Sweeney, L. (1997). Guaranteeing anonymity when sharing medical data, the Datafly System. In Proceedings of the AMIA Annual Fall Symposium (p. 51). American Medical Informatics Association

    1997

  45. [53]

    K., Chundawat, V

    Tarun, A. K., Chundawat, V. S., Mandal, M., & Kankanhalli, M. (2023). Fast yet effective machine unlearning. IEEE Transactions on Neural Networks and Learning Systems

    2023

  46. [54]

    F., Kieseberg, P., & Li, T

    Villaronga, E. F., Kieseberg, P., & Li, T. (2018). Humans forget, machines remember: Artificial intelligence and the right to be forgotten. Computer Law & Security Review, 34(2), 304-313

    2018

  47. [55]

    Wang, Z., Zheng, Z., Jiang, W., & Tang, S. (2021). Blockchain‐enabled data sharing in supply chains: Model, operationalization, and tutorial. Production and Operations Management, 30(7), 1965-1985

    2021

  48. [56]

    Xu, H., & Dinev, T. (2022). Guest Editorial: Reflections on the 2021 Impact Award: Why Privacy Still Matters. MIS Quarterly, 46(4)

    2022

  49. [57]

    J., Chen, D., Chau, M., Li, L., & Zheng, H

    Xu, J. J., Chen, D., Chau, M., Li, L., & Zheng, H. (2022). Peer-to-peer loan fraud detection: constructing features from transaction data. MIS quarterly, 46(3)

    2022

  50. [58]

    Xu, H., & Zhang, N. (2022a). From contextualizing to context theorizing: assessing context effects in privacy research. Management Science, 68(10), 7383-7401

    2022

  51. [59]

    Xu, H., & Zhang, N. (2022b). Implications of data anonymization on the statistical evidence of disparity. Management Science, 68(4), 2600-2618

    2022

  52. [60]

    Xu, H., Zhu, T., Zhang, L., Zhou, W., & Yu, P. S. (2023). Machine unlearning: A survey. ACM Computing Surveys, 56(1), 1-36

    2023

  53. [61]

    Yu, S., Chai, Y., Chen, H., Sherman, S., & Brown, R. (2021). Wearable sensor-based chronic condition severity assessment: An adversarial attention-based deep multisource multitask learning approach. Forthcoming in MIS Quarterly

    2021

  54. [62]

    Zhang, D., Finckenberg-Broman, P., Hoang, T., Pan, S., Xing, Z., Staples, M., & Xu, X. (2023). Right to be forgotten in the era of large language models: Implications, challenges, and solutions. arXiv preprint arXiv:2307.03941

    2023 arXiv

  55. [63]

    & Gao, X

    Zhou, J., Li, H., Liao, X., Zhang, B., He, W., Li, Z., ... & Gao, X. (2023). A unified method to revoke the private data of patients in intelligent healthcare with audit to forget. Nature Communications, 14(1), 6255

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.