Beyond Synthetic Augmentation: Group-Aware Threshold Calibration for Robust Balanced Accuracy in Imbalanced Learning

Hunter Gittlin

REVIEW 4 major objections 6 minor 16 references

Beyond Synthetic Augmentation: Group-Aware Threshold Calibration for Robust Balanced Accuracy in Imbalanced Learning

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper claims that in imbalanced classification with protected groups, calibrating a separate decision threshold for each group outperforms synthetic oversampling, making augmentation largely redundant.

desk verdict A transparent but over-claimed empirical study: the '1.5-4% better' headline compares tuned thresholds to untuned baselines, and a fair tuned-vs-tuned reading shows the advantage mostly evaporates. read the letter →

arxiv 2509.02592 v1 pith:A26MHF77 submitted 2025-08-29 cs.LG cs.AI

Hunter Gittlin This is my paper

classification cs.LGcs.AI

keywords classimbalancebalancedaccuracyworst-groupgroup-awarethresholdcalibrationsyntheticoversamplingSMOTECT-GANalgorithmicfairness

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's central claim is that when class imbalance and protected groups coincide, adjusting the decision threshold separately for each group is a more direct and more robust fix than generating synthetic training data. Analyzing two financial benchmarks, seven model families, and a grid search over group-specific thresholds, the author reports that group thresholds on unmodified data beat SMOTE and CT-GAN augmented pipelines by 1.5-4% in balanced accuracy while also improving worst-group balanced accuracy. The key evidence for redundancy is that adding group thresholds to synthetic data produces only small extra gains over thresholds on original data. If the claim holds, standard oversampling pipelines for imbalanced group-sensitive problems may be unnecessary complexity, and auditors gain an interpretable lever over group-level error rates.

What carries the argument

The load-bearing object is the group-aware threshold, a per-protected-group cutoff τ_g applied to predicted probabilities so that an instance is classified positive when p_i ≥ τ_g(i). The thresholds are selected with a validation-set grid search under two objectives: Fair-BalAcc, which maximizes overall balanced accuracy, and Fair-MinBalAcc, which maximizes worst-group balanced accuracy. This machinery does the argument's work because it directly targets the evaluation metric, keeps the original training distribution intact, and converts group-level fairness into an explicit, interpretable control, while the comparison baselines (SMOTE and CT-GAN) act as the competing mechanisms that turn ou

What would settle it

Run the same comparison on an extreme-imbalance dataset (e.g., 1% positive class) with a continuous or multi-label protected attribute. If SMOTE or CT-GAN plus group thresholds improves balanced accuracy or worst-group accuracy over original-data thresholds by more than the 1.5-4% gap seen here, the redundancy claim is false. A sharper version: for any augmented model, check whether there exists a per-group threshold on the original model's scores that matches or beats the augmented model's balanced accuracy; if not, augmentation adds information thresholds cannot recover.

Watch

Extended reading notes

Core claim

The author sets out to establish that synthetic augmentation (SMOTE interpolation and CT-GAN conditional generation) and threshold-based class-imbalance remedies are not complementary: they are redundant, because both aim at the same underlying metric, and the threshold route dominates. The evidence is a systematic comparison on UCI Default of Credit Card Clients (sex as the protected group) and Adult Income (race as the protected group), across logistic regression, SGD, random forest, histogram gradient boosting, XGBoost, CatBoost, and k-NN. For every model family, group-specific thresholds fit on original data achieve the highest or near-highest balanced accuracy and worst-group balanced a

Load-bearing premise

The claim of fundamental redundancy rests on two moderate-imbalance, binary-attribute datasets being representative; if extreme imbalance, multi-class outcomes, continuous protected attributes, or interacting groups behave differently, the conclusion collapses.

Editorial extensions

If this is right

Practitioners facing binary class imbalance with a protected attribute can skip SMOTE/CT-GAN preprocessing: group-thresholding the original data gives 1.5-4% higher balanced accuracy and higher worst-group balanced accuracy across linear, tree, boosting, and instance-based models.
Thresholding synthetic data yields little beyond thresholding original data, so the two techniques should not be combined expecting additive gains; the computational budget for augmentation is better spent on validation-set threshold search.
Using the Fair-MinBalAcc objective gives a direct control on the worst-off group's balanced accuracy, exposing a Pareto frontier instead of a single accuracy number.
The results support using protected attributes as a calibration input rather than excluding them, a point with direct relevance to fairness regulation.
Because the benefit appears across seven model families, the conclusion is not an artifact of one classifier family.

Reading between the lines

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

The paper's two datasets both sit near 4:1 imbalance with one binary protected attribute; the natural next test is whether extreme imbalance (e.g., 100:1) or multiple intersecting groups breaks the redundancy pattern. Until that is run, the 'fundamentally redundant' conclusion should be read as bounded to the tested regime.
A direct corollary the author does not spell out: the diminishing-return pattern gives practitioners a cheap diagnostic. If the threshold gain on augmented data is close to zero, the augmentation is not earning its cost; if it is large, augmentation is adding information.
The redundancy claim suggests a sharper identity to test: any balanced-accuracy gain from augmentation should be reproducible by some per-group threshold on original-model scores. Proving or disproving that identity would turn the empirical result into a general theorem.
If regulators read this result, the policy implication is to regulate how sensitive attributes are used in calibration rather than ban them outright; a flat ban may forgo the exact group-accuracy control the method demonstrates.

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.

Desk Editor's Note

read the letter

Dear colleague,

The one thing you should know: this paper is a clean, honest empirical comparison, but its headline claim doesn't survive a fair comparison. The '1.5–4% higher balanced accuracy' in the abstract is computed against raw SMOTE/CT-GAN rows, which appear to use a default 0.5 threshold. The paper's own tuned-vs-tuned rows (Original+Fair-BalAcc vs SMOTE+Fair-BalAcc) show the advantage shrinks to about 1–2 percentage points on the credit data and frequently to zero or negative on Adult. So the stated magnitude and consistency aren't supported by the paper's own numbers.

What is genuinely useful here is the redundancy observation: applying group-aware thresholds to already-augmented data gives little extra benefit. That's a practical data point, and it aligns with the literature the author cites (GHOST, Wallace et al., Hardt et al.). The experimental setup is transparent: five-fold CV, seven model families, group-specific threshold search via OxonFair, and both balanced accuracy and worst-group balanced accuracy are reported. The paper is honest about its scope, explicitly conceding it only tests binary outcomes, binary protected attributes, and moderate imbalance.

The soft spots are more than cosmetic. There are no standard deviations or significance tests, so even the small tuned-vs-tuned gaps on the credit data could be noise. The validation split used for threshold selection is 12.5% of training, and the reported numbers don't account for the selection-on-validation bias. And while the conclusions are couched in the language of 'fundamental redundancy,' that's a stretch for two datasets. The citation pattern is fine—the prior work is acknowledged, and the incremental contribution is clearly framed as an empirical check rather than a new mechanism.

For whom is this useful? Anyone who wants a quick sanity check that threshold tuning beats oversampling, especially in fairness-constrained settings. It won't change practice on its own. It deserves a serious referee only if the venue is willing to ask for a revised version with tuned-vs-tuned comparisons and error bars; as is, the abstract overstates what the tables show.

My recommendation: don't rely on the headline numbers. If you're reading this area, it's worth twenty minutes, but I wouldn't cite it in its current form.

Best.

Referee Report

4 major / 6 minor

Summary. The paper proposes group-aware threshold calibration—learning separate decision thresholds for each protected demographic group—as an alternative to synthetic oversampling methods (SMOTE, CT-GAN) for class-imbalanced data. Experiments on two financial benchmarks (UCI Default of Credit Card Clients, Adult Income) with seven model families compare original, SMOTE-augmented, and CT-GAN-augmented training sets under raw and threshold-tuned evaluation, reporting balanced accuracy (BA) and worst-group balanced accuracy (WG-BA). The paper claims that group-aware thresholds on original data achieve 1.5–4% higher balanced accuracy than SMOTE/CT-GAN augmented models, improve worst-group performance, and that applying the same thresholds to synthetic data adds little, implying synthetic augmentation and threshold calibration are 'fundamentally redundant.' The central conclusion is that practitioners should first apply group-aware threshold calibration to original data before considering synthetic augmentation.

Significance. If the claims were fully supported, the contribution would be practically valuable: a simple, interpretable, computationally cheap post-processing method that improves both balanced accuracy and worst-group balanced accuracy without synthetic data. The experimental design has strengths: it spans seven model families, reports both BA and WG-BA, applies group-aware threshold selection on a held-out validation split (avoiding circularity), and uses an informative incremental-gain comparison (thresholds on original vs. thresholds on synthetic data). The paper is also transparent enough that the reported tables permit independent re-analysis. However, as presented, the headline effect size and the 'consistently outperforms' claim rest on comparing tuned thresholds against untuned raw baselines, and no uncertainty quantification is provided. The conceptual finding that synthetic augmentation adds little once thresholds are tuned may survive a fair re-analysis, but the current evidence does not establish the stated magnitude, consistency, or generality.

major comments (4)

[§4.1, Tables 1–2, Abstract] The abstract's '1.5–4% higher balanced accuracy' and Section 4.1's narrative compare Original+Fair-BalAcc with SMOTE-Raw and CTGAN-Raw rows, i.e., a tuned method against default-threshold baselines. Recomputing the tuned-vs-tuned gap (Original Fair-BalAcc minus SMOTE/CT-GAN Fair-BalAcc) across the 14 model-dataset rows gives means of about 1.3 pp for SMOTE and 0.6 pp for CT-GAN, with several rows below 0.3 pp or negative (e.g., k-NN on Adult: -0.4 pp vs. SMOTE; CatBoost on Credit: -0.1 pp vs. CT-GAN; CatBoost on Adult: 0.0 pp vs. CT-GAN). The largest tuned-vs-tuned advantage is about 3.4 pp, not 4%. The stated magnitude and 'consistently outperforms' are therefore not supported by a fair comparison. The manuscript should either make the tuned-vs-tuned comparison primary or explicitly scope the claim to default-threshold practitioners.
[§3.1, §3.3, Tables 1–2] All results are reported as point estimates from 5-fold stratified cross-validation, with no standard deviations, confidence intervals, or significance tests. Once the comparison is made tuned-vs-tuned, most differences are at or below 1 percentage point—well within typical fold-to-fold variation for datasets of 30,000 and 48,842 instances. The phrase in §4.1 that the pattern 'holds remarkably consistent' is not verifiable without uncertainty information. Report per-fold results, repeated-run variability, or at least a paired significance test for the key comparisons that support the abstract's claims.
[§3.1, §5.3, Abstract/Conclusion] The categorical conclusion that synthetic augmentation and threshold calibration are 'fundamentally redundant' is extrapolated from two datasets, both with binary protected attributes and moderate (~4:1) imbalance. The manuscript's own Limitations section (§5.3) concedes that extreme imbalance, multi-class imbalance, continuous protected attributes, and multi-group settings are untested. The wording in the title, abstract, and conclusion overstates the evidence. Please restrict the conclusion to the tested regime or add experiments supporting generality; otherwise the central claim is a claim about a much broader class of problems than the data can support.
[§3.1, §2.2] The CT-GAN baseline is trained for only 5 epochs, with no generation-quality diagnostics or hyperparameter search reported. The paper uses CT-GAN's raw performance as a key comparator in §4.1 and concludes in §2.2 that CT-GAN 'fails to improve' despite its computational complexity. An undertrained generator could make the synthetic baseline artificially weak. Report training convergence, generated-sample quality, or additional epochs, and at minimum justify why the chosen CT-GAN configuration is a representative strong baseline.

minor comments (6)

[§2.3, Eq. (2)] The indicator notation '⊮' in Eq. (2) is nonstandard; use a standard indicator (e.g., 1[·]) and define g(i). Also, §2.3 contains a typo: 'a approach' should be 'an approach.'
[Table 2 caption] Table 2's caption says 'Adult Income dataset ... on the credit-default dataset.' This is a typo; the dataset is Adult Income, not credit-default. Also, the caption repeats 'credit-default' while Table 1 is the credit data.
[§1 and §5.1] There is a sentence fragment in §1: 'As these synthetic samples often create overlapping class regions...' should be merged with the preceding sentence. §5.1 ends with an extra period after 'groups.'.
[§4.2] The sentence 'supporting [14] observation' should read 'supporting [14]'s observation' or 'supporting the observation of [14].'
[§5.2 and Conclusion] The claim that the method provides 'orders of magnitude less computation' is not measured or cited; it should be removed or supported with wall-clock measurements.
[General] No code, data, or random seeds are provided. Given that the headline claims rest on small differences, releasing code and per-fold results would substantially strengthen reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular dependency: threshold calibration is validated on held-out splits and compared against external baselines; the headline overstatement is a fairness-of-comparison issue, not circularity.

full rationale

The paper's derivation chain is purely empirical and self-contained. Group-aware thresholds are selected on a reserved validation split (Section 3.1: 'Within training data, we reserve 12.5% for threshold optimization using OxonFair's grid search over group-specific thresholds, ensuring no data leakage'), and all BA/WG-BA results are measured on held-out test predictions. No equation defines a reported result in terms of a fitted parameter or in terms of the conclusion itself. The 'redundancy' argument in Section 4.2 compares the incremental gain from thresholding original data against the incremental gain from thresholding SMOTE/CT-GAN data; that is an empirical comparison, not a definitional reduction. The paper contains no self-citations, so there is no self-citation chain or imported uniqueness theorem. The limitation that raw SMOTE/CT-GAN baselines are untuned while Original+Fair-BalAcc is tuned is a legitimate experimental-design criticism: it makes the headline 1.5-4% figure appear larger than the tuned-vs-tuned differences in Tables 1 and 2. However, unfair comparison or interpretive overreach is not circularity, because the underlying comparisons are externally falsifiable and the results are not forced by construction. Section 5.3 explicitly concedes untested regimes (multi-class, continuous protected attributes, extreme imbalance), which limits generalizability but does not make the argument circular. Overall, no load-bearing circular step exists.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claim rests on metric choice (balanced accuracy), test-time group membership, and dataset representativeness. No new physical or mathematical entities are introduced, and no fitted constants beyond standard hyperparameters appear. The main unstated costs are the threshold grid search choices and the single-dataset-regime generalization.

free parameters (3)

Threshold grid resolution = not specified
OxonFair's grid search over group-specific thresholds is the core fitting step, but the grid granularity and range are not reported, making the threshold choice a hidden tuning parameter.
CT-GAN training epochs = 5
The generative model is trained for only 5 epochs (Section 3.1), a hand-chosen value that may understate CT-GAN's potential and affects the comparison.
Validation split fraction = 12.5% of training data
The split used for threshold selection is fixed at 12.5% (Section 3.1), which influences threshold estimates and is a modeling choice.

assumptions (3)

domain assumption Balanced accuracy is the appropriate target metric for imbalanced learning.
Section 3.3 defines BA as the primary metric; the entire comparison assumes BA is what practitioners should optimize, which is reasonable but still a value-laden modeling choice.
domain assumption Protected group membership is known at deployment time for every instance.
Equation (2) applies group-specific thresholds using g(i), requiring group labels at test time. In many regulatory contexts this is contested or unavailable.
domain assumption Two tabular datasets with binary protected attributes and moderate ~4:1 imbalance are representative of imbalanced learning with protected groups.
Section 3.1 selects UCI Default and Adult Income; Section 5.3 admits extreme imbalance, multi-class, and continuous attributes are untested, yet the conclusions are stated generally.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Beyond Synthetic Augmentation: Group-Aware Threshold Calibration for Robust Balanced Accuracy in Imbalanced Learning." pith.science (2026). https://pith.science/paper/A26MHF77

@misc{pith2026250902592,
  author       = {Pith},
  title        = {Pith review of: Beyond Synthetic Augmentation: Group-Aware Threshold Calibration for Robust Balanced Accuracy in Imbalanced Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A26MHF77}},
  note         = {Machine review of arXiv:2509.02592}
}

read the original abstract

Class imbalance remains a fundamental challenge in machine learning, with traditional solutions often creating as many problems as they solve. We demonstrate that group-aware threshold calibration--setting different decision thresholds for different demographic groups--provides superior robustness compared to synthetic data generation methods. Through extensive experiments, we show that group-specific thresholds achieve 1.5-4% higher balanced accuracy than SMOTE and CT-GAN augmented models while improving worst-group balanced accuracy. Unlike single-threshold approaches that apply one cutoff across all groups, our group-aware method optimizes the Pareto frontier between balanced accuracy and worst-group balanced accuracy, enabling fine-grained control over group-level performance. Critically, we find that applying group thresholds to synthetically augmented data yields minimal additional benefit, suggesting these approaches are fundamentally redundant. Our results span seven model families including linear, tree-based, instance-based, and boosting methods, confirming that group-aware threshold calibration offers a simpler, more interpretable, and more effective solution to class imbalance.

Figures

Figures reproduced from arXiv: 2509.02592 by the authors.

**Figure 1.** Left: Classification of male and female non-defaulters and defaulters in a single threshold setting. This single threshold performs poorly, misclassifying three out of seven individuals (roughly 57% accuracy). Right: Classification of male and female non-defaulters and defaulters in a group thresholding scenario, in which there is a separate threshold for males and females. In this case, the model correctly classifi… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

16 extracted references · 4 canonical work pages

[1]

BMC Bioinformatics 14(1), 106 (Dec 2013)

Blagus, R., Lusa, L.: SMOTE for high-dimensional class-imbalanced data. BMC Bioinformatics 14(1), 106 (Dec 2013). https://doi.org/10.1186/1471-2105-14-106, https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-14-106

doi:10.1186/1471-2105-14-106 2013
[2]

Journal of Artificial Intelligence Research16, 321–357 (Jun 2002)

Chawla, N.V., Bowyer, K.W., Hall, L.O., Kegelmeyer, W.P.: SMOTE: Synthetic Minority Over-sampling Technique. Journal of Artificial Intelligence Research16, 321–357 (Jun 2002). https://doi.org/10.1613/jair.953, http://arxiv.org/abs/1106. 1813, arXiv:1106.1813 [cs]

arXiv 2002
[3]

https://doi.org/10.48550/arXiv.1701

Corbett-Davies, S., Pierson, E., Feller, A., Goel, S., Huq, A.: Algorithmic decision making and the cost of fairness (Jun 2017). https://doi.org/10.48550/arXiv.1701. 08230, http://arxiv.org/abs/1701.08230, arXiv:1701.08230 10 Hunter Gittlin( )

arXiv doi:10.48550/arxiv.1701.08230 2017
[4]

https://doi.org/10.48550/arXiv.2407

Delaney, E., Fu, Z., Wachter, S., Mittelstadt, B., Russell, C.: OxonFair: A Flexible Toolkit for Algorithmic Fairness (Nov 2024). https://doi.org/10.48550/arXiv.2407. 13710, http://arxiv.org/abs/2407.13710, arXiv:2407.13710 [cs]

arXiv doi:10.48550/arxiv.2407.13710 2024
[5]

https://doi.org/10.48550/ arXiv.2008.09202, http://arxiv.org/abs/2008.09202, arXiv:2008.09202 [cs]

Engelmann, J., Lessmann, S.: Conditional Wasserstein GAN-based Oversampling of Tabular Data for Imbalanced Learning (Aug 2020). https://doi.org/10.48550/ arXiv.2008.09202, http://arxiv.org/abs/2008.09202, arXiv:2008.09202 [cs]

arXiv doi:10.48550/arxiv.2008.09202 2020
[6]

Journal of Chemical Information and Modeling61(6), 2623–2640 (Jun 2021)

Esposito, C., Landrum, G.A., Schneider, N., Stiefl, N., Riniker, S.: GHOST: Adjust- ing the Decision Threshold to Handle Imbalanced Data in Machine Learning. Journal of Chemical Information and Modeling61(6), 2623–2640 (Jun 2021). https://doi. org/10.1021/acs.jcim.1c00160, https://pubs.acs.org/doi/10.1021/acs.jcim.1c00160

doi:10.1021/acs.jcim.1c00160 2021
[7]

IEEE Transactions on Knowledge and Data Engineering21(9), 1263–1284 (Sep 2009)

Haibo He, Garcia, E.: Learning from Imbalanced Data. IEEE Transactions on Knowledge and Data Engineering21(9), 1263–1284 (Sep 2009). https://doi.org/10. 1109/TKDE.2008.239, http://ieeexplore.ieee.org/document/5128907/

work page arXiv 2009
[8]

https://doi.org/10.48550/arXiv.1610.02413, http://arxiv.org/abs/1610

Hardt, M., Price, E., Srebro, N.: Equality of Opportunity in Supervised Learning (Oct 2016). https://doi.org/10.48550/arXiv.1610.02413, http://arxiv.org/abs/1610. 02413, arXiv:1610.02413

arXiv doi:10.48550/arxiv.1610.02413 2016

Show all 16 references

[9]

https://doi

Hassanat, A.B., Tarawneh, A.S., Altarawneh, G.A., Almuhaimeed, A.: Stop Over- sampling for Class Imbalance Learning: A Critical Review (Jun 2022). https://doi. org/10.48550/arXiv.2202.03579, http://arxiv.org/abs/2202.03579, arXiv:2202.03579 [cs]

2022 doi arXiv
[10]

Applied Soft Computing83, 105662 (Oct 2019)

Kovács, G.: An empirical comparison and evaluation of minority oversampling techniques on a large number of imbalanced datasets. Applied Soft Computing83, 105662 (Oct 2019). https://doi.org/10.1016/j.asoc.2019.105662, https://linkinghub. elsevier.com/retrieve/pii/S1568494619304429

2019
[11]

https://doi.org/10.48550/arXiv.2107.09044, http: //arxiv.org/abs/2107.09044, arXiv:2107.09044 [cs]

Liu, E.Z., Haghgoo, B., Chen, A.S., Raghunathan, A., Koh, P.W., Sagawa, S., Liang, P., Finn, C.: Just Train Twice: Improving Group Robustness without Training Group Information (Sep 2021). https://doi.org/10.48550/arXiv.2107.09044, http: //arxiv.org/abs/2107.09044, arXiv:2107....

2021 doi arXiv
[12]

Pattern Recognition 91, 216–231 (Jul 2019)

Luque, A., Carrasco, A., Martín, A., De Las Heras, A.: The impact of class imbalance in classification performance metrics based on the binary confusion matrix. Pattern Recognition 91, 216–231 (Jul 2019). https://doi.org/10.1016/j.patcog.2019.02.023, https://linkinghub.elsevie...

2019 doi
[13]

https://doi.org/10.48550/arXiv.1911.08731, http:// arxiv.org/abs/1911.08731, arXiv:1911.08731 [cs]

Sagawa, S., Koh, P.W., Hashimoto, T.B., Liang, P.: Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization (Apr 2020). https://doi.org/10.48550/arXiv.1911.08731, http:// arxiv.org/abs/1911.08731, arXiv:1911.08731 [cs]

2020 doi arXiv
[14]

IEEE Computational Intelligence Magazine 13(4), 59–76 (Nov 2018)

Santos, M.S., Soares, J.P., Abreu, P.H., Araujo, H., Santos, J.: Cross-Validation for Imbalanced Datasets: Avoiding Overoptimistic and Overfitting Approaches [Research Frontier]. IEEE Computational Intelligence Magazine 13(4), 59–76 (Nov 2018). https://doi.org/10.1109/MCI.2018...

2018
[15]

In: 2011 IEEE 11th International Conference on Data Mining

Wallace, B.C., Small, K., Brodley, C.E., Trikalinos, T.A.: Class Imbalance, Redux. In: 2011 IEEE 11th International Conference on Data Mining. pp. 754–763. IEEE, Vancouver, BC, Canada (Dec 2011). https://doi.org/10.1109/ICDM.2011.33, http: //ieeexplore.ieee.org/document/6137280/

2011
[16]

https://doi.org/10.48550/arXiv.1907

Xu, L., Skoularidou, M., Cuesta-Infante, A., Veeramachaneni, K.: Modeling Tabular data using Conditional GAN (Oct 2019). https://doi.org/10.48550/arXiv.1907. 00503, http://arxiv.org/abs/1907.00503, arXiv:1907.00503 [cs]

2019 doi arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

BMC Bioinformatics 14(1), 106 (Dec 2013)

Blagus, R., Lusa, L.: SMOTE for high-dimensional class-imbalanced data. BMC Bioinformatics 14(1), 106 (Dec 2013). https://doi.org/10.1186/1471-2105-14-106, https://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-14-106

doi:10.1186/1471-2105-14-106 2013

[2] [2]

Journal of Artificial Intelligence Research16, 321–357 (Jun 2002)

Chawla, N.V., Bowyer, K.W., Hall, L.O., Kegelmeyer, W.P.: SMOTE: Synthetic Minority Over-sampling Technique. Journal of Artificial Intelligence Research16, 321–357 (Jun 2002). https://doi.org/10.1613/jair.953, http://arxiv.org/abs/1106. 1813, arXiv:1106.1813 [cs]

arXiv 2002

[3] [3]

https://doi.org/10.48550/arXiv.1701

Corbett-Davies, S., Pierson, E., Feller, A., Goel, S., Huq, A.: Algorithmic decision making and the cost of fairness (Jun 2017). https://doi.org/10.48550/arXiv.1701. 08230, http://arxiv.org/abs/1701.08230, arXiv:1701.08230 10 Hunter Gittlin( )

arXiv doi:10.48550/arxiv.1701.08230 2017

[4] [4]

https://doi.org/10.48550/arXiv.2407

Delaney, E., Fu, Z., Wachter, S., Mittelstadt, B., Russell, C.: OxonFair: A Flexible Toolkit for Algorithmic Fairness (Nov 2024). https://doi.org/10.48550/arXiv.2407. 13710, http://arxiv.org/abs/2407.13710, arXiv:2407.13710 [cs]

arXiv doi:10.48550/arxiv.2407.13710 2024

[5] [5]

https://doi.org/10.48550/ arXiv.2008.09202, http://arxiv.org/abs/2008.09202, arXiv:2008.09202 [cs]

Engelmann, J., Lessmann, S.: Conditional Wasserstein GAN-based Oversampling of Tabular Data for Imbalanced Learning (Aug 2020). https://doi.org/10.48550/ arXiv.2008.09202, http://arxiv.org/abs/2008.09202, arXiv:2008.09202 [cs]

arXiv doi:10.48550/arxiv.2008.09202 2020

[6] [6]

Journal of Chemical Information and Modeling61(6), 2623–2640 (Jun 2021)

Esposito, C., Landrum, G.A., Schneider, N., Stiefl, N., Riniker, S.: GHOST: Adjust- ing the Decision Threshold to Handle Imbalanced Data in Machine Learning. Journal of Chemical Information and Modeling61(6), 2623–2640 (Jun 2021). https://doi. org/10.1021/acs.jcim.1c00160, https://pubs.acs.org/doi/10.1021/acs.jcim.1c00160

doi:10.1021/acs.jcim.1c00160 2021

[7] [7]

IEEE Transactions on Knowledge and Data Engineering21(9), 1263–1284 (Sep 2009)

Haibo He, Garcia, E.: Learning from Imbalanced Data. IEEE Transactions on Knowledge and Data Engineering21(9), 1263–1284 (Sep 2009). https://doi.org/10. 1109/TKDE.2008.239, http://ieeexplore.ieee.org/document/5128907/

work page arXiv 2009

[8] [8]

https://doi.org/10.48550/arXiv.1610.02413, http://arxiv.org/abs/1610

Hardt, M., Price, E., Srebro, N.: Equality of Opportunity in Supervised Learning (Oct 2016). https://doi.org/10.48550/arXiv.1610.02413, http://arxiv.org/abs/1610. 02413, arXiv:1610.02413

arXiv doi:10.48550/arxiv.1610.02413 2016