Pith. sign in

REVIEW 3 major objections 3 minor 57 references

Regression Augmentation With Data-Driven Segmentation

T0 review · 3 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Data-driven minority segmentation beats fixed-threshold augmentation in imbalanced regression.

desk verdict A genuinely data-driven segmentation idea for imbalanced regression, backed by a broad empirical comparison; the main risk is uncorrected multiple-testing in the significant-win counts. read the letter →

arxiv 2508.01455 v1 pith:HF5NOR6S submitted 2025-08-02 cs.LG

Shayan Alahyari , Shiva Mehdipour Ghobadlou , Mike Domaratzki This is my paper
classification cs.LG
keywords imbalancedregressiondataaugmentationGANGaussianmixturemodelMahalanobisdistancenearest-neighbourmatchingoversamplingbenchmarkevaluation
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

Imbalanced regression arises when a continuous target has a skewed distribution, so standard regressors overfit dense regions and make large errors on rare extremes. The paper claims that fixed thresholds for labelling samples as rare overlook the structure of the joint feature–target space, and that a fully data-driven alternative can do better. It proposes an augmentation framework: a Gaussian mixture model with Mahalanobis distance identifies minority samples, a GAN generates synthetic rare examples, and deterministic nearest-neighbour matching places them in sparse regions. Across 32 imbalanced regression benchmarks, the paper reports consistent improvements over random oversampling, SMOGN, and G-SMOTE on RMSE, SERA, and F-measure, using both TabNet and Random Forest regressors. If correct, this means imbalanced regression need not rely on ad hoc rarity thresholds; the data itself can point to the regions that need enrichment.

What carries the argument

Three components carry the argument. First, Mahalanobis–GMM segmentation: a Gaussian mixture model is fitted on the joint feature–target space, and the Mahalanobis distance $\sqrt{(x-\mu)^\top \Sigma^{-1}(x-\mu)}$ from a sample to its assigned component determines whether it is rare, replacing a fixed target threshold with a covariance-aware boundary. Second, a GAN trained on the labelled rare samples generates synthetic minority examples. Third, deterministic nearest-neighbour matching pairs each generated sample with the nearest real sample in the joint space, keeping the augmentation inside genuinely sparse regions. The Mahalanobis distance is the piece that lets the method adapt to the local shape of the data, which a univariate target-value cut cannot.

What would settle it

Take a synthetic regression dataset with a known rare region that is strongly non-Gaussian (for example, a thin spiral or two disconnected islands in the target's sparse tail). Run the proposed method and compare its labelled rare set with the ground-truth rare region. If the overlap is close to zero, or if the augmented regressor fails to beat a fixed-threshold baseline on SERA restricted to that region, the claim that data-driven segmentation identifies true minority regions is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that data-driven minority segmentation, rather than preset thresholds, is what makes regression augmentation work. The method fits a Gaussian mixture model over the joint feature–target space, uses the Mahalanobis distance to each mixture component to label samples as rare or common, trains a GAN on the rare samples, and then uses deterministic nearest-neighbour matching to anchor each synthetic sample to a real neighbour. The reported result is that this pipeline consistently outperforms random oversampling, SMOGN, and G-SMOTE across RMSE, SERA, and F-measure on 32 benchmark datasets, for both TabNet and Random Forest regressors. The authors take this as evidence that adversarial generation combined with distance-based matching produces synthetic samples that align with the true distribution of rare targets.

Load-bearing premise

The load-bearing premise is that a Gaussian mixture model on the joint feature–target space, combined with Mahalanobis distance, correctly identifies which samples are truly rare; if the mixture misses a rare cluster or the covariance estimate becomes unstable in a small sparse sample, the GAN will enrich the wrong regions and the reported improvement collapses.

Editorial extensions

If this is right

  • Users of imbalanced regression no longer need to set a rarity threshold; the GMM discovers which regions are sparse from the data.
  • The gain is not tied to one model: the framework improves both TabNet and Random Forest on the 32 benchmarks.
  • SERA improvements indicate that the largest gains appear exactly on the rare targets, where standard regressors fail.
  • The combination of GAN synthesis with nearest-neighbour anchoring suggests a template for other augmentation pipelines: generate with a model, then snap back to the original distribution.

Reading between the lines

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

  • The learned rarity labels could be used with any oversampler, not only a GAN; swapping SMOGN or G-SMOTE into the same segmentation would isolate how much of the gain comes from the segmentation step versus the generator.
  • A stress test with a non-Gaussian rare region (a thin manifold, a ring, or two disconnected islands) would show whether the GMM's covariance assumption is the binding constraint on the data-driven claim.
  • In streaming settings where the target skew drifts, the GMM can be refit online to re-discover minority regions, something a fixed-threshold method cannot do.
  • The deterministic matching step offers a cheap diagnostic: if the synthetic samples' Mahalanobis distances to the real rare distribution grow large, the generator is drifting and the matching step is masking it.
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

3 major / 3 minor

Summary. The paper proposes a GAN-based data augmentation framework for imbalanced regression. The method uses Mahalanobis-Gaussian Mixture Modeling (GMM) to automatically identify minority samples in the joint feature-target space without preset thresholds, and then applies deterministic nearest-neighbour matching to generate synthetic samples for those sparse regions. The authors claim that, evaluated on 32 benchmark imbalanced regression datasets, the proposed method consistently outperforms random oversampling, SMOGN, G-SMOTE, and a no-oversampling baseline across RMSE, SERA, and F-measure, with statistical significance assessed by Wilcoxon signed-rank tests. The visible evidence consists of aggregate win counts and significance results for two regressors (TabNet and Random Forest) reported in Figures 8 and 9.

Significance. If the empirical claims hold, this is a useful contribution to imbalanced regression, especially the fully data-driven segmentation mechanism that avoids ad hoc thresholds. The evaluation across 32 datasets and two different regressors is a commendable breadth. However, the credibility of the paper rests almost entirely on the experimental protocol and the reported significance tests, and the provided text lacks sufficient detail to assess reproducibility or to verify that the significance claims are not inflated by multiple comparisons. The method's reliance on GMM-based segmentation also introduces a modeling assumption that deserves explicit robustness analysis.

major comments (3)
  1. [Figures 8 and 9 and the surrounding text] The 'statistically significant wins' are reported at an alpha level of 0.05 with no correction for multiple comparisons. With three metrics and at least four baselines, the paper performs at least 12 separate Wilcoxon signed-rank tests per regressor; under the global null, one would expect about 0.6 false positives per regressor by chance alone. The headline claim that the proposed method 'secures the majority of statistically significant wins' could therefore be inflated. Please report adjusted p-values (e.g., Holm-Bonferroni or false discovery rate) or provide a justification for why multiple-testing correction is unnecessary in this setting.
  2. [Method description and experimental setup] The manuscript text provided does not include the full method details or experimental protocol: it omits the number of GMM components, the Mahalanobis distance threshold or quantile used to label minority samples, the GAN architecture and training hyperparameters, the value of k in the nearest-neighbour matching, the list of the 32 datasets, and any measures of variance across runs. Without these details, the empirical claim cannot be reproduced or independently verified. Please include a complete experimental setup section with exact settings and dataset characteristics.
  3. [Data-driven segmentation (GMM step)] The proposed method's rationale is that the GMM over the joint feature-target space, combined with Mahalanobis distance, correctly identifies the true rare regions. This is a load-bearing assumption, because if the mixture model underfits a rare cluster or the covariance estimate becomes unstable in small sparse samples, the GAN will enrich the wrong regions. The paper reports no sensitivity analysis with respect to the number of GMM components or the Mahalanobis threshold. Please add a robustness study varying these hyperparameters and report whether the performance gains are stable across reasonable choices.
minor comments (3)
  1. [Abstract] In the sentence 'Existing approaches often rely on fixed, ad hoc thresholds... and fail to represent the true underlying rare regions', the subject 'approaches' does not agree with the verb 'fail'; consider revising to '...and they often fail to represent...'.
  2. [Figures 8 and 9] The bar charts show only aggregate win counts without any indication of variability or confidence intervals. Adding error bars or a matrix of per-metric, per-baseline results would make the comparison more informative and would allow readers to assess the consistency of the wins.
  3. [Statistical testing description] The text should state precisely what is being compared in each Wilcoxon signed-rank test: whether the test is over the 32 datasets for each metric separately, and how the 'wins' are defined when multiple runs are performed. This would clarify the interpretation of the reported p-values.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; the empirical comparison is grounded in external baselines and standard fitting procedures.

full rationale

The visible manuscript describes an empirical augmentation framework: a GMM with Mahalanobis distance is fit to identify minority samples, a GAN generates synthetic data, and nearest-neighbour matching enriches sparse regions. The performance claim is then evaluated against external baselines (random oversampling, SMOGN, G-SMOTE, and a no-oversampling control) on 32 benchmark datasets using RMSE, SERA, and F-measure. No equation in the visible text defines a predicted quantity in terms of a fitted parameter in a way that would make the result true by construction. The GMM and GAN are fit on training data, which is ordinary model fitting rather than circular reasoning. The cited works (Krawczyk, Branco, Torgo, Chawla, Guo, Johnson, Wilcoxon) are background and methodological references, not self-citations carrying the main claim. The reported Wilcoxon signed-rank tests at alpha 0.05 without visible multiple-testing correction raise a statistical-significance concern, but that is a correctness risk, not circularity. The paper's central comparison is externally grounded, so the circularity score is 0.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

No code or full method details are available from the abstract and partial excerpt. The framework introduces no new physical entities, but it depends on several modeling choices and hyperparameters that are not disclosed in the visible text. These choices directly affect which samples are augmented and therefore the quality of the empirical claim.

free parameters (4)
  • GMM number of components = not reported
    The abstract does not state how the number of mixture components is selected; this is a free choice that can change which regions are labeled rare.
  • Mahalanobis distance threshold or quantile for minority labeling = not reported
    The method uses Mahalanobis distance to identify minority samples, but the exact cutoff that separates rare from common is not specified.
  • GAN architecture and training hyperparameters = not reported
    GAN training is sensitive to architecture, learning rate, and stopping criteria, none of which are mentioned in the abstract.
  • Nearest-neighbour matching parameter k = not reported
    The deterministic nearest-neighbour matching step requires a choice of k or a distance threshold, which is not specified.
assumptions (3)
  • domain assumption Synthetic oversampling of rare target values improves regression performance on rare regions without hurting dense regions.
    The entire method assumes that generating extra examples in sparse target regions will improve the regression model, an assumption that is plausible but not guaranteed.
  • domain assumption A Gaussian mixture model on the joint feature-target space accurately captures the true underlying rare regions.
    The abstract states the GMM is used to automatically identify minority samples. This assumes the data within each cluster is approximately Gaussian and that sparse clusters correspond to meaningful rare regions.
  • domain assumption Mahalanobis distance computed from estimated cluster covariance is a valid measure of rarity in the feature-target space.
    Mahalanobis distance assumes a reliable covariance estimate, which can be unstable when rare clusters are very small.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Regression Augmentation With Data-Driven Segmentation." pith.science (2026). https://pith.science/paper/HF5NOR6S

@misc{pith2026250801455,
  author       = {Pith},
  title        = {Pith review of: Regression Augmentation With Data-Driven Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HF5NOR6S}},
  note         = {Machine review of arXiv:2508.01455}
}
read the original abstract

Imbalanced regression arises when the target distribution is skewed, causing models to focus on dense regions and struggle with underrepresented (minority) samples. Despite its relevance across many applications, few methods have been designed specifically for this challenge. Existing approaches often rely on fixed, ad hoc thresholds to label samples as rare or common, overlooking the continuous complexity of the joint feature-target space and fail to represent the true underlying rare regions. To address these limitations, we propose a fully data-driven GAN-based augmentation framework that uses Mahalanobis-Gaussian Mixture Modeling (GMM) to automatically identify minority samples and employs deterministic nearest-neighbour matching to enrich sparse regions. Rather than preset thresholds, our method lets the data determine which observations are truly rare. Evaluation on 32 benchmark imbalanced regression datasets demonstrates that our approach consistently outperforms state-of-the-art data augmentation methods.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

57 extracted references · 49 canonical work pages

  1. [1]

    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 bbl.in ":" * " " * FUNCTION f...

  2. [2]

    An introduction to outlier analysis

    Charu C Aggarwal. An introduction to outlier analysis. Springer, 2017

    work page 2017

  3. [3]

    Local distribution-based adaptive oversampling for imbalanced regression

    Shayan Alahyari and Mike Domaratzki. Local distribution-based adaptive oversampling for imbalanced regression. Preprint arXiv:2504.14316, 2025 a

    work page Pith review arXiv 2025

  4. [4]

    SMOGAN: Synthetic Minority Oversampling with GAN Refinement for Imbalanced Regression

    Shayan Alahyari and Mike Domaratzki. SMOGAN : Synthetic minority oversampling with GAN refinement for imbalanced regression. Preprint arXiv:2504.21152, 2025 b

    work page Pith review arXiv 2025

  5. [5]

    Alcalá-Fdez, A

    J. Alcalá-Fdez, A. Fernandez, J. Luengo, J. Derrac, S. García, L. Sánchez, and F. Herrera. Keel data-mining software tool: Data set repository, integration of algorithms and experimental analysis framework. Journal of Multiple-Valued Logic and Soft Computing, 17 0 (2-3): 0 255--287, 2011

    work page 2011

  6. [6]

    Ar k and Tomas Pfister

    Sercan \"O . Ar k and Tomas Pfister. Tabnet: Attentive interpretable tabular learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 6679--6687, 2021. doi:10.1609/aaai.v35i8.16826

    doi:10.1609/aaai.v35i8.16826 2021

  7. [7]

    Wasserstein GAN

    Martin Arjovsky, Soumith Chintala, and Léon Bottou. Wasserstein GAN . arXiv preprint arXiv:1701.07875v3 [stat.ML], 2017

    arXiv 2017

  8. [8]

    Mmote: A Mahalanobis distance-based oversampling approach for improving minority class classification in parkinson’s disease detection

    Kavitha Balakrishnan and Anitha Shenoy. Mmote: A Mahalanobis distance-based oversampling approach for improving minority class classification in parkinson’s disease detection. Applied Soft Computing, 136: 0 110174, 2023. doi:10.1016/j.asoc.2023.110174

    work page arXiv 2023

Show all 57 references
  1. [9]

    Branco, L

    P. Branco, L. Torgo, and R. P. Ribeiro. A survey of predictive modeling under imbalanced distributions. ACM Computing Surveys, 49 0 (2): 0 Article 31, 2016

    2016

  2. [10]

    Branco, L

    P. Branco, L. Torgo, and R. P. Ribeiro. SMOGN : A pre-processing approach for imbalanced regression. In Proceedings of Machine Learning Research: LIDTA, volume 74, pages 36--50, 2017

    2017

  3. [11]

    Branco, L

    P. Branco, L. Torgo, and R. P. Ribeiro. Pre-processing approaches for imbalanced distributions in regression. Neurocomputing, 343: 0 76--99, 2019

    2019

  4. [12]

    Random forests

    Leo Breiman. Random forests. Machine Learning, 45 0 (1): 0 5--32, 2001. doi:10.1023/A:1010933404324

    2001 doi

  5. [13]

    M. Buda, A. Maki, and M. A. Mazurowski. A systematic study of the class imbalance problem in convolutional neural networks. Neural Networks, 106: 0 249--259, 2018

    2018

  6. [14]

    Camacho and F

    L. Camacho and F. Bacao. WSMOTER : A novel approach for imbalanced regression. Applied Intelligence, 54: 0 8789--8799, 2024

    2024

  7. [15]

    Camacho, G

    L. Camacho, G. Douzas, and F. Bacao. Geometric SMOTE for regression. Expert Systems with Applications, 193: 0 116387, 2022

    2022

  8. [16]

    N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer. SMOTE : Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16: 0 321--357, 2002

    2002

  9. [17]

    N. V. Chawla, N. Japkowicz, and A. Kolcz. Editorial: Special issue on learning from imbalanced data sets. ACM SIGKDD Explorations Newsletter, 6 0 (1): 0 1--6, 2004

    2004

  10. [18]

    Domingos

    P. Domingos. Metacost: A general method for making classifiers cost-sensitive. In Proceedings of the 5th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), pages 155--164, 1999

    1999

  11. [19]

    C. Elkan. The foundations of cost-sensitive learning. In Proceedings of the 17th International Joint Conference on Artificial Intelligence (IJCAI), pages 973--978, 2001

    2001

  12. [20]

    Conditional Wasserstein GAN ‐based oversampling of tabular data for imbalanced learning

    Justin Engelmann and Stefan Lessmann. Conditional Wasserstein GAN ‐based oversampling of tabular data for imbalanced learning. Preprint arXiv:2008.09202v1, 2020

    2008 arXiv

  13. [21]

    Mahalanobis distance and its application for detecting multivariate outliers

    Hamid Ghorbani. Mahalanobis distance and its application for detecting multivariate outliers. Facta Universitatis, Series: Mathematics and Informatics, pages 583--595, 2019

    2019

  14. [22]

    Deep Learning

    Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep Learning. MIT Press, 2016. http://www.deeplearningbook.org

    2016

  15. [23]

    Improved training of Wasserstein GANs

    Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron Courville. Improved training of Wasserstein GANs . In Advances in Neural Information Processing Systems 30 (NeurIPS 2017), pages 5767--5777, 2017

    2017

  16. [24]

    H. Guo, Y. Li, J. Shang, M. Gu, Y. Huang, and B. Gong. Learning from class-imbalanced data: Review of methods and applications. Expert Systems with Applications, 73: 0 220--239, 2017

    2017

  17. [25]

    He and E

    H. He and E. A. Garcia. Learning from imbalanced data. IEEE Transactions on Knowledge and Data Engineering, 21 0 (9): 0 1263--1284, 2009

    2009

  18. [26]

    A GAN ‐based anomaly detection approach for imbalanced industrial time series

    Wenqian Jiang, Yang Hong, Beitong Zhou, Xin He, and Cheng Cheng. A GAN ‐based anomaly detection approach for imbalanced industrial time series. IEEE Access, 7: 0 143608--143619, 2019

    2019

  19. [27]

    J. M. Johnson and T. M. Khoshgoftaar. Survey on deep learning with class imbalance. Journal of Big Data, 6 0 (1): 0 1--54, 2019

    2019

  20. [28]

    Sams, B.\, Nuno Dokoozlian, Luis Sanchez, and M

    Hossein Kamangir, S. Sams, B.\, Nuno Dokoozlian, Luis Sanchez, and M. Earles, J.\. Large-scale spatio-temporal yield estimation via deep learning using satellite and management data fusion in vineyards. Computers and Electronics in Agriculture, 216: 0 108439, 2024

    2024

  21. [29]

    Krawczyk

    B. Krawczyk. Learning from imbalanced data: open challenges and future directions. Progress in Artificial Intelligence, 5 0 (4): 0 221--232, 2016

    2016

  22. [30]

    N. Kunz. Smogn: Synthetic minority over-sampling technique for regression with gaussian noise. PyPI, version v0.1.2, 2020

    2020

  23. [31]

    Gan-based imbalanced data intrusion detection system

    JooHwa Lee and KeeHyun Park. Gan-based imbalanced data intrusion detection system. Personal and Ubiquitous Computing, 25 0 (1): 0 121--128, 2021

    2021

  24. [32]

    X.-Y. Liu, J. Wu, and Z.-H. Zhou. Exploratory undersampling for class-imbalance learning. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 39 0 (2): 0 539--550, 2009

    2009

  25. [33]

    Yihong Ma, Xiaobao Huang, Bozhao Nan, Nuno Moniz, Xiangliang Zhang, Olaf Wiest, and Nitesh V. Chawla. Are we making much progress? revisiting chemical reaction yield prediction from an imbalanced regression perspective. In Companion Proceedings of the ACM Web Conference 2024 (...

    2024

  26. [34]

    Bryan F. J. Manly. Multivariate Statistical Methods: A Primer. Chapman & Hall/CRC Press, Boca Raton, FL ; London, 3rd edition, 2005

    2005

  27. [35]

    BAGAN : Data augmentation with balancing GAN

    Giovanni Mariani, Florian Scheidegger, Roxana Istrate, Costas Bekas, and Cristiano Malossi. BAGAN : Data augmentation with balancing GAN . Preprint arXiv:1803.09655v2, 2018

    2018 arXiv

  28. [36]

    The EM algorithm and extensions

    Geoffrey J McLachlan and Thriyambakam Krishnan. The EM algorithm and extensions. John Wiley & Sons, 2008

    2008

  29. [37]

    Moniz, L

    N. Moniz, L. Torgo, and C. Soares. SMOTEBoost for regression: Improving the prediction of extreme values. In Proceedings of the 5th International Conference on Data Science and Advanced Analytics (DSAA), pages 127--136, 2018

    2018

  30. [38]

    Software defect prediction using over-sampling and feature extraction based on mahalanobis distance

    Mohammad Mahdi NezhadShokouhi, Mohammad Ali Majidi, and Abbas Rasoolzadegan. Software defect prediction using over-sampling and feature extraction based on mahalanobis distance. The Journal of Supercomputing, 76 0 (1): 0 602--635, 2020. doi:10.1007/s11227-019-03051-w

    2020 doi

  31. [39]

    M. Ren, W. Luo, and R. Urtasun. Balanced mse for imbalanced visual regression. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 418--427, 2022

    2022

  32. [40]

    R. P. Ribeiro and N. Moniz. Imbalanced regression and extreme value prediction. Machine Learning, 109 0 (9-10): 0 1803--1835, 2020

    2020

  33. [41]

    R. P. A. Ribeiro. Utility-based regression. PhD thesis, Faculty of Sciences, University of Porto, Porto, 2011

    2011

  34. [42]

    On the use of the Gaussian mixture model and the Mahalanobis distance for fault diagnosis in dynamic components of electric motors

    Ronny Francis Ribeiro Junior, Fabricio Alves de Almeida, Ariosto Bretanha Jorge, João Luiz Junho Pereira, Matheus Brendon Francisco, and Guilherme Ferreira Gomes. On the use of the Gaussian mixture model and the Mahalanobis distance for fault diagnosis in dynamic components of...

    2023 doi

  35. [43]

    Scheepens, D.\, R

    R. Scheepens, D.\, R. Müller, J. Vos, I. Giannakas, E. Portilla-Figueroa, L. Prieto, and F. Correoso. Adapting a deep convolutional recurrent neural network model for improved spatio-temporal forecasting of extreme wind speed events using imbalanced regression losses. Geoscien...

    2023

  36. [44]

    SMOTified - GAN for class imbalanced pattern classification problems

    Anuraganand Sharma, Prabhat Kumar Singh, and Rohitash Chandra. SMOTified - GAN for class imbalanced pattern classification problems. IEEE Access, 10: 0 30655--30665, 2022

    2022

  37. [45]

    Steininger, K

    M. Steininger, K. Kobs, P. Davidson, A. Krause, and A. Hotho. Density-based weighting for imbalanced regression. Machine Learning, 110 0 (8): 0 2187--2210, 2021

    2021

  38. [46]

    Data augmentation using GANs

    Fabio Henrique Kiyoiti dos Santos Tanaka and Claus Aranha. Data augmentation using GANs . In Proceedings of Machine Learning Research, volume XXX, pages 1--16, 2019

    2019

  39. [47]

    Torgo, R

    L. Torgo, R. P. Ribeiro, J. P. da Costa, and S. Pal. SMOTE for regression. In Intelligent Data Engineering and Automated Learning (IDEAL 2013). Lecture Notes in Computer Science, volume 8206, pages 378--387, 2013

    2013

  40. [48]

    Utility-based regression

    Luis Torgo and Rita Ribeiro. Utility-based regression. In Proceedings of the 11th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD 2007), pages 597--604, 2007

    2007

  41. [49]

    Precision and recall for regression

    Luis Torgo and Rita Ribeiro. Precision and recall for regression. In Discovery Science (DS 2009), volume 5808 of Lecture Notes in Artificial Intelligence, pages 332--346. Springer‑Verlag Berlin Heidelberg, 2009

    2009

  42. [50]

    Visualizing data using t-sne

    Laurens van der Maaten and Geoffrey Hinton. Visualizing data using t-sne. Journal of Machine Learning Research, 9: 0 2579--2605, 2008

    2008

  43. [51]

    Gaussian mixture modeling by exploiting the mahalanobis distance

    Dimitrios Ververidis and Constantine Kotropoulos. Gaussian mixture modeling by exploiting the mahalanobis distance. IEEE Transactions on Signal Processing, 56 0 (7): 0 2797--2807, July 2008. doi:10.1109/TSP.2008.917350

    2008

  44. [52]

    Wilcoxon

    F. Wilcoxon. Individual comparisons by ranking methods. Biometrics Bulletin, 1 0 (6): 0 80--83, 1945

    1945

  45. [53]

    W. Wu, N. Kunz, and P. Branco. Imbalancedlearningregression-a python package to tackle the imbalanced regression problem. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pages 645--648, 2022

    2022

  46. [54]

    J. Yang, L. Xie, Q. Yu, X. He, and J. Liu. Delving into deep imbalanced regression. In Proceedings of the 38th International Conference on Machine Learning (ICML), pages 8437--8447, 2021

    2021

  47. [55]

    Evolutionary mahalanobis distance-based oversampling for multi-class imbalanced data classification

    Leehter Yao and Tung-Bin Lin. Evolutionary mahalanobis distance-based oversampling for multi-class imbalanced data classification. Sensors, 21 0 (19): 0 6616, 2021. doi:10.3390/s21196616

    2021 doi

  48. [56]

    Phan, and Peng Chench

    Qian Zhang, Yue Wang, Li Chen, Hao Li, Thanh N. Phan, and Peng Chench. Irmae-akde: A novel deep imbalanced regression approach for performance prediction of rolled steel plate yield strength. IEEE Transactions on Instrumentation and Measurement, 74: 0 2006911, 2025

    2025

  49. [57]

    Zhou and X.-Y

    Z.-H. Zhou and X.-Y. Liu. On multi-class cost-sensitive learning. Computational Intelligence, 26 0 (3): 0 232--257, 2010

    2010

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.