Deep Learning Meets Oversampling: A Learning Framework to Handle Imbalanced Classification

Asela Hevapathige; Sukumar Kishanthan

REVIEW 2 major objections 5 minor 50 references

Deep Learning Meets Oversampling: A Learning Framework to Handle Imbalanced Classification

T0 review · 2 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read The paper claims that oversampling for imbalanced classification can be reformulated as a composition of learnable discrete decision criteria, packaged as an end-to-end deep framework (AutoSMOTE) whose cohort variant achieves the best…

desk verdict A genuinely novel oversampling mechanism, but the test-set hyperparameter tuning and the circular VC-dimension 'proof' sink the paper's central claims. read the letter →

arxiv 2502.06878 v1 pith:MSFEJPQV submitted 2025-02-08 cs.LG

Sukumar Kishanthan , Asela Hevapathige This is my paper · ORCID

classification cs.LG

keywords classimbalanceoversamplingdeeplearningmulti-layerperceptronrepresentationGumbel-SoftmaxVCdimensionsyntheticminority

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 argues that oversampling done as a separate preprocessing step is mismatched with the downstream classifier, because synthetic samples are generated without regard to the loss that the model will be optimized for. It introduces AutoSMOTE, a deep learning framework that folds oversampling into training: for each minority instance, the model learns whether to oversample it, how many nearest neighbors to use, and which aggregation function to apply. These discrete choices are made differentiable with Gumbel-Softmax, so the sampling policy and the classifier are trained jointly by backpropagation. The paper reports that this decision-based formulation improves generalization relative to plain MLP oversamplers, and that AutoSMOTEcohort ranks first overall on eight imbalanced datasets.

What carries the argument

The load-bearing mechanism is a decision criteria mapping ψj = μj ∘ Φj, where Φj maps a minority instance to an integer and μj maps that integer to one of finitely many predefined decisions. For each of the three criteria—oversampling participation, k-nearest-neighbor count, and aggregation function—an MLP produces class probabilities, and Gumbel-Softmax converts them into differentiable one-hot choices. The chosen decisions are composed to generate a synthetic sample: if participation is selected, the chosen k neighbors are combined by the chosen function, drawn from linear interpolation, minimum, maximum, sum, average, and weighted average. This composition defines the hypothesis classes whose nesting (cohort ⊂ self ⊂ MLP) drives the VC-dimension generalization argument.

What would settle it

Train AutoSMOTEcohort, AutoSMOTEself, and an MLP oversampler to matched training error on an imbalanced dataset and compare held-out test error; if the MLP has lower test error, or if the ordering is not cohort < self < MLP, the claimed generalization hierarchy is contradicted.

Watch

Extended reading notes

Core claim

The central discovery is that an oversampling function is better modeled as a composition of discrete decision criteria than as a free-form continuous function. AutoSMOTEself makes the three decisions per minority instance independently, while AutoSMOTEcohort first learns a grouping of minority instances and applies the decision criteria to each group collectively. Both variants are trained end to end with an MLP classifier under categorical cross-entropy, with Gumbel-Softmax supplying gradients through the discrete selections. The paper's theoretical claim is that the resulting hypothesis classes are nested—AutoSMOTEcohort inside AutoSMOTEself inside the MLP oversampler—so their VC dimensions (a standard capacity measure) and hence their generalization errors are ordered accordingly, which the authors then validate empirically.

Load-bearing premise

The load-bearing premise is that the nested hypothesis classes have VC dimensions in the same order and that generalization error scales with VC dimension; the paper asserts both steps as reasonable expectations rather than proving them.

Editorial extensions

If this is right

If the generalization ordering holds, AutoSMOTEcohort should consistently beat AutoSMOTEself and MLP oversamplers on held-out test sets, not only on the eight datasets reported.
Because the oversampling policy is trained jointly with the classifier, the framework can be attached to any differentiable classifier without redesigning the sampling step.
The restricted hypothesis class lowers the risk of overfitting in small minority classes, which matters for high-dimensional or sparse tabular datasets.
The learned aggregation functions allow the model to switch between SMOTE-style interpolation and nonlinear summaries per instance, making the synthesis policy more flexible than a fixed SMOTE.

Reading between the lines

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

Editorial inference: the same decision-criteria composition could be applied to undersampling or cost-sensitive weighting, since the mechanism only requires differentiable selection over discrete choices.
Editorial inference: because participation, k, and aggregation are interpretable, the learned policy can be inspected to see which synthetic strategy the model favors in different minority regions, giving a post-hoc explanation of when a dataset benefits from interpolation versus extremal aggregation.
Editorial inference: a natural extension is to test whether the relative importance of the three criteria shifts with imbalance ratio; the paper's ablation covers only three datasets, so the pattern across a wider range of skews remains an open question.

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

The core idea here is real: a learnable, end-to-end oversampler where discrete choices—which minority instances to use, how many neighbors, which aggregation function—are made by Gumbel-Softmax MLPs jointly trained with the classifier. That is a legitimate twist on OC-SMOTE-NN and worth exploring. The two variants are coherent and the ablation study asks the right questions. But the paper's two load-bearing pillars do not hold up.

First, the experimental protocol is invalid. Section 5.4 says the data are split once into 80% training and 20% testing, and then each model is run across 2 to 6 nearest neighbors “and record the best performance metric.” For AutoSMOTEcohort, the number of groups is searched over 1 to 7 and the best recorded. There is no validation set. Every entry in Tables 1–2 and Figure 3 is therefore the product of selecting hyperparameters on the test split, and the cohort variant has extra knobs to overfit with. The reported superiority is not supported. Second, Theorem 2 is not a proof. It assumes the hypothesis class nesting, asserts without derivation that the VC dimensions follow the same order, and then asserts ϵgen ∝ VCdim to conclude the generalization ordering. Section 6.1's “validation” is a qualitative look at training/test curves and does not rescue it.

I want to give credit where it is due: the framework is clearly explained, the related work is covered reasonably well, and the authors are not sloppy writers. The problems are deeper than presentation. Some smaller issues also bother me: missing ADASYN results for several datasets, identical standard deviations across different methods in places, and no statistical significance testing anywhere. The claim that MLPs have higher interpretability than GANs is at least debatable and could use support.

Bottom line: this is a salvageable paper, not a publishable one as-is. The mechanism is novel and the flaws are fixable. A proper revision would use a validation-based tuning protocol, either produce a real generalization bound or drop the theory, and add proper statistical comparisons. If this crossed my desk, I would send it to reviewers with that explicit framing rather than desk-reject it, because the core idea deserves referee time. But I would not cite it in its current form, and I would not hold it up as a model of experimental practice.

Referee Report

2 major / 5 minor

Summary. The paper proposes AutoSMOTE, a deep-learning-based oversampling framework for imbalanced classification. The method formulates oversampling as a composition of discrete decision criteria (oversampling participation, k-nearest-neighbor count, and aggregation function), learned end-to-end with MLPs and a Gumbel-Softmax relaxation. Two variants are introduced, AutoSMOTEself (per-instance decisions) and AutoSMOTEcohort (decisions applied to learnable groups of minority instances). The manuscript claims a theoretical generalization bound via VC dimension that ranks the two variants below a plain MLP in generalization error, and it reports experiments on eight tabular datasets claiming that AutoSMOTEcohort achieves the best overall average rank and lower test error than MLP-based oversampling.

Significance. The core idea of treating oversampling as a set of learnable discrete decisions that are jointly optimized with the classifier is genuinely interesting and could be a useful contribution to imbalanced classification. The paper also provides an ablation study separating the effect of each decision criterion and reports a runtime comparison, which are valuable elements. However, the central empirical claim of superiority is invalidated by the model-selection protocol, and the theoretical result is not actually proven. If the method were evaluated with a proper validation-based hyperparameter selection and the theoretical claim were either rigorously established or removed, the underlying framework could be of interest to the community, but in its current form the paper's headline conclusions are not supported.

major comments (2)

[5.4] The hyperparameter selection protocol leaks test information. Section 5.4 states that 'we run each model across 2 to 6 nearest neighbors and record the best performance metric' and for AutoSMOTEcohort 'we perform a search across 1 to 7 groups and record the best result,' but no validation set is described; the data is split once into 80% training and 20% testing. Therefore the numbers in Tables 1, 2, and 3 are selected on the test split rather than being unbiased estimates of generalization. AutoSMOTEcohort has an additional tunable parameter (the number of groups), so it has more opportunity to benefit from this test-set overfitting, directly undermining the claim in Table 2 that it achieves the best overall rank and the Section 6.1 claim that it has lower test error than the MLP oversampler. This is a load-bearing flaw because the main contribution is empirical superiority.
[4, Theorem 2] The proof of Theorem 2 is not a derivation. It assumes without proof that HAutoSMOTEcohort ⊂ HAutoSMOTEself ⊂ HMLP, then asserts 'it is reasonable to expect that their VC dimensions would follow the same order,' and finally uses an asserted proportionality ϵgen(H) ∝ VCdim(H). None of these steps is established: hypothesis class inclusion is not formally shown from the architectural description; VC dimension does not in general follow from class inclusion; and Definition 4 provides an upper bound involving sqrt(VCdim log N/N), not a proportionality of the generalization error itself. The theorem's conclusion therefore rests on the very ordering it purports to prove. Section 6.1 claims to empirically validate the theorem, but the empirical comparison is already invalidated by the test-set selection issue in Section 5.4, and training error curves (Figure 2) do not directly demonstrate a VC-dimension ordering.

minor comments (5)

[3.5.2] The variant names are inconsistently and incorrectly typeset as 'AutoSM OT Eself' and 'AutoSM OT Ecohort' in Section 3.5.2, which should be cleaned up throughout.
[5.1] There are grammatical issues such as 'It has of 5473 data instances' (Page-blocks), 'It contains 2417 data instances with 103 attributes' (Yeast) where 'attributes' is likely 'features,' and 'datasets related for diverse fields' in the opening sentence of Section 5.1.
[2] The phrase 'MPLs' appears in the Related Work section ('MPLs have higher interpretability'), apparently a typo for 'MLPs.'
[Table 1, Wisconsin] For the Wisconsin dataset, the reported Precision, Recall, and F1-score for AutoSMOTEself and AutoSMOTEcohort are identical to three decimal places, which is suspicious and likely a copy-paste error; the authors should confirm which values are correct.
[6.1] Figures 2 and 3 are referenced but not visible in the manuscript text provided for review; the authors should ensure the figures are included and clearly labeled, and ideally also report the numerical values of training and test error so the claimed trends can be assessed.

Circularity Check

2 steps flagged · score 6.0 of 10

Superiority claim is supported by a test-set selection protocol and a theorem whose proof assumes its own ordering; reported best results are fitted rather than predicted.

fitted input called prediction [Section 5.4 (Model Hyper-parameters); results in Section 6.0.1, Tables 1 and 2]
"Therefore, we run each model across 2 to 6 nearest neighbors and record the best performance metric, enabling fair comparison. Further, AutoSMOT Ecohort requires determining the optimal number of groups into which minority samples should be grouped. In order to do that, we perform a search across 1 to 7 groups and record the best result."

Section 5.4 describes only an 80:20 training/testing split and no validation split. Thus 'record the best performance metric' means taking the maximum over test-set evaluations for each k, and for AutoSMOTEcohort it also means maximizing over the group count 1-7. The reported best overall rank in Table 2 is therefore the result of selecting hyperparameters on the test data. The claim that AutoSMOTEcohort 'emerges the best performing model' is not an out-of-sample prediction; it is the value of the test-set objective used to choose the model. The extra group-count search gives AutoSMOTEcohort additional selection opportunities, so its reported superiority is forced by the protocol rather than independently demonstrated.
other [Section 4, Theorem 2 and its proof]
"From the design, we know that HAutoSMOTEcohort ⊂ HAutoSMOTEself ⊂ HMLP. Therefore, it is reasonable to expect that their VC-dimensions would follow the same order: VC dim(HAutoSMOTEcohort ) < VCdim(HAutoSMOTEself ) < VCdim(HMLP). ... Under the given assumptions, we can approximate the following relationship: ϵgen(H) ∝ VCdim(H). This approximation would leads to the inequality: ϵgen(HAutoSMOTEcohort ) < ϵgen(HAutoSMOTEself ) < ϵgen(HMLP). This completes the proof."

The two load-bearing statements in the proof are the VC-dimension ordering and the proportionality ϵgen(H) ∝ VCdim(H); neither is derived from the model definitions or from earlier theorems. Since the proportionality makes the generalization ordering a relabeling of the VC-dimension ordering, the theorem's conclusion is the assumed ordering expressed in different notation. The subsequent sentence 'we empirically validate these theoretical findings in Section 6.1' closes the loop using the test-selected results of Section 5.4, so the theoretical and empirical supports reduce to the same unproven ordering.

full rationale

The paper's central empirical claim, that AutoSMOTEcohort has the best overall average rank, is not an independent estimate: Section 5.4 tunes k in 2-6 and the cohort group count in 1-7 by recording the best performance metric on the test split, with no validation set described. Because AutoSMOTEcohort has the extra group-count hyperparameter, its top rank in Table 2 is partly an artifact of maximizing over the test data; this is a fitted input presented as a prediction. Section 6.1's confirmation of Theorem 2 inherits the same test-set bias. The theoretical Theorem 2 is also not an independent derivation: its proof assumes the VC-dimension ordering and assumes ϵgen ∝ VCdim, so the conclusion is the assumed ordering restated rather than derived from established facts. No load-bearing self-citation chain exists: the self-citations ([15], [30]) provide background and a baseline or formalism, but they do not carry the uniqueness or validity of the proposed design. Score 6 reflects that a central 'prediction' (empirical superiority) reduces by construction to the selection protocol, with partial theoretical circularity in Theorem 2.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central theoretical claim rests on several unproven assumptions about hypothesis class nesting and VC dimension ordering, and the empirical claim rests on hyperparameters tuned on the test set.

free parameters (4)

K-nearest neighbors k for oversampling baselines = selected as best over range 2-6 on test set
Section 5.4: 'we run each model across 2 to 6 nearest neighbors and record the best performance metric'. This is a free parameter chosen to maximize reported test performance.
Number of groups for AutoSMOTEcohort = selected as best over range 1-7 on test set
Section 5.4: 'perform a search across 1 to 7 groups and record the best result'. This hyperparameter is tuned on the test set.
Batch size = chosen from {500, 2500, 5000}
Section 5.4: batch size is selected for each dataset, a hand-chosen hyperparameter affecting minority sampling.
Gumbel-Softmax temperature tau = not reported
Equation (6) defines tau but the paper never specifies its value or schedule, leaving a free parameter unstated.

assumptions (6)

standard math Universal approximation theorem for MLPs
Invoked in Section 4 to justify MLP-based decision criteria.
standard math VC dimension generalization bound ϵgen ≤ ϵtrain + O(sqrt(VCdim log N / N))
Definition 4 in Section 4.
ad hoc to paper Hypothesis class nesting HAutoSMOTEcohort ⊂ HAutoSMOTEself ⊂ HMLP
Theorem 2 proof assumes this nesting 'from the design' without formal construction.
ad hoc to paper VC dimensions follow the same order as hypothesis class nesting
'It is reasonable to expect that their VC-dimensions would follow the same order' (Theorem 2 proof), an asserted rather than proven premise.
ad hoc to paper Proportionality ϵgen(H) ∝ VCdim(H)
'We can approximate the following relationship: ϵgen(H) ∝ VCdim(H)', which is not a standard result and is used to complete the proof.
domain assumption Training errors of the three models are approximately equal
Stated as condition 1 in Theorem 2; asserted rather than established.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Deep Learning Meets Oversampling: A Learning Framework to Handle Imbalanced Classification." pith.science (2026). https://pith.science/paper/MSFEJPQV

@misc{pith2026250206878,
  author       = {Pith},
  title        = {Pith review of: Deep Learning Meets Oversampling: A Learning Framework to Handle Imbalanced Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MSFEJPQV}},
  note         = {Machine review of arXiv:2502.06878}
}

read the original abstract

Despite extensive research spanning several decades, class imbalance is still considered a profound difficulty for both machine learning and deep learning models. While data oversampling is the foremost technique to address this issue, traditional sampling techniques are often decoupled from the training phase of the predictive model, resulting in suboptimal representations. To address this, we propose a novel learning framework that can generate synthetic data instances in a data-driven manner. The proposed framework formulates the oversampling process as a composition of discrete decision criteria, thereby enhancing the representation power of the model's learning process. Extensive experiments on the imbalanced classification task demonstrate the superiority of our framework over state-of-the-art algorithms.

Figures

Figures reproduced from arXiv: 2502.06878 by the authors.

**Figure 2.** Training Error Comparison of AutoSMOTE with MLP-Oversampler [PITH_FULL_IMAGE:figures/full_fig_p025_2.png] view at source ↗

**Figure 3.** Testing Performance Comparison of AutoSMOTE with MLP-Oversampler. [PITH_FULL_IMAGE:figures/full_fig_p026_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Training Time Comparison for Oversampling Algorithms [PITH_FULL_IMAGE:figures/full_fig_p028_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 45 canonical work pages

[1]

X. Guo, Y. Yin, C. Dong, G. Yang, G. Zhou, On the class imbalance problem, in: 2008 Fourth international conference on natural computa- tion, Vol. 4, IEEE, 2008, pp. 192–201

work page 2008
[2]

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

work page 2019
[3]

D. A. Cieslak, N. V. Chawla, A. Striegel, Combating imbalance in net- work intrusion datasets., in: GrC, 2006, pp. 732–737

work page 2006
[4]

Japkowicz, S

N. Japkowicz, S. Stephen, The class imbalance problem: A systematic study, Intelligent data analysis 6 (5) (2002) 429–449

2002
[5]

K. G. Al-Hashedi, P. Magalingam, Financial fraud detection applying data mining techniques: A comprehensive review from 2009 to 2019, Computer Science Review 40 (2021) 100402

work page 2021
[6]

Khushi, K

M. Khushi, K. Shaukat, T. M. Alam, I. A. Hameed, S. Uddin, S. Luo, X. Yang, M. C. Reyes, A comparative performance analysis of data resampling methods on imbalance medical data, IEEE Access 9 (2021) 109960–109975

work page 2021
[7]

A. I. Marqu´ es, V. Garc ´ ıa, J. S. S´ anchez, On the suitability of resampling techniques for the class imbalance problem in credit scoring, Journal of the Operational Research Society 64 (2013) 1060–1070

work page 2013
[8]

N. V. Chawla, K. W. Bowyer, L. O. Hall, W. P. Kegelmeyer, Smote: 32 synthetic minority over-sampling technique, Journal of artificial intelli- gence research 16 (2002) 321–357

work page 2002

Show all 50 references

[9]

Tang, Y.-Q

Y. Tang, Y.-Q. Zhang, N. V. Chawla, S. Krasser, Svms modeling for highly imbalanced classification, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) 39 (1) (2008) 281–288

2008
[10]

Han, W.-Y

H. Han, W.-Y. Wang, B.-H. Mao, Borderline-smote: a new over- sampling method in imbalanced data sets learning, in: International conference on intelligent computing, Springer, 2005, pp. 878–887

2005
[11]

H. He, Y. Bai, E. A. Garcia, S. Li, Adasyn: Adaptive synthetic sampling approach for imbalanced learning, in: 2008 IEEE international joint conference on neural networks (IEEE world congress on computational intelligence), Ieee, 2008, pp. 1322–1328

2008
[12]

Douzas, F

G. Douzas, F. Bacao, F. Last, Improving imbalanced learning through a heuristic oversampling method based on k-means and smote, Informa- tion sciences 465 (2018) 1–20

2018
[13]

S. Ando, C. Y. Huang, Deep over-sampling framework for classifying imbalanced data, in: Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2017, Skopje, Mace- donia, September 18–22, 2017, Proceedings, Part I 10, Springer, 2017, pp. 770–785

2017
[14]

Dablain, B

D. Dablain, B. Krawczyk, N. V. Chawla, Deepsmote: Fusing deep learn- ing and smote for imbalanced data, IEEE Transactions on Neural Net- works and Learning Systems. 33
[15]

Karunasingha, B

N. Karunasingha, B. G. Jayasekara, A. Hevapathige, Oc-smote-nn: A deep learning-based approach for imbalanced classification, in: 2023 IEEE 13th Annual Computing and Communication Workshop and Con- ference (CCWC), IEEE, 2023, pp. 0943–0948

2023
[16]

Sharma, A

S. Sharma, A. Gosain, S. Jain, A review of the oversampling techniques in class imbalance problem, in: International Conference on Innovative Computing and Communications: Proceedings of ICICC 2021, Volume 1, Springer, 2022, pp. 459–472

2021
[17]

R. A. Sowah, B. Kuditchar, G. A. Mills, A. Acakpovi, R. A. Twum, G. Buah, R. Agboyi, Hcbst: An efficient hybrid sampling technique for class imbalance problems, ACM Transactions on Knowledge Discovery from Data (TKDD) 16 (3) (2021) 1–37

2021
[18]

Gosain, S

A. Gosain, S. Sardana, Handling class imbalance problem using over- sampling techniques: A review, in: 2017 international conference on ad- vances in computing, communications and informatics (ICACCI), IEEE, 2017, pp. 79–85

2017
[19]

D. Devi, S. K. Biswas, B. Purkayastha, A review on solution to class imbalance problem: Undersampling approaches, in: 2020 international conference on computational performance evaluation (ComPE), IEEE, 2020, pp. 626–631

2020
[20]

C. X. Ling, V. S. Sheng, Cost-sensitive learning and the class imbalance problem, Encyclopedia of machine learning 2011 (2008) 231–235. 34

2008
[21]

Tyagi, S

S. Tyagi, S. Mittal, Sampling approaches for imbalanced data classi- fication problem in machine learning, in: Proceedings of ICRIC 2019: Recent innovations in computing, Springer, 2020, pp. 209–221

2019
[22]

Fern´ andez, S

A. Fern´ andez, S. Garc ´ ıa, M. Galar, R. C. Prati, B. Krawczyk, F. Herrera, A. Fern´ andez, S. Garc ´ ıa, M. Galar, R. C. Prati, et al., Algorithm-level approaches, Learning from Imbalanced Data Sets (2018) 123–146

2018
[23]

Ahmed, S

Z. Ahmed, S. Das, A comparative analysis on recent methods for ad- dressing imbalance classification, SN Computer Science 5 (1) (2023) 30

2023
[24]

F. Last, G. Douzas, F. Bacao, Oversampling for imbalanced learning based on k-means and smote arXiv:1711.00837

arXiv
[25]

S. S. Mullick, S. Datta, S. Das, Generative adversarial minority over- sampling, in: Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1695–1704

2019
[26]

Engelmann, S

J. Engelmann, S. Lessmann, Conditional wasserstein gan-based over- sampling of tabular data for imbalanced learning, Expert Systems with Applications 174 (2021) 114582

2021
[27]

Troullinou, G

E. Troullinou, G. Tsagkatakis, A. Losonczy, P. Poirazi, P. Tsakalides, A generative neighborhood-based deep autoencoder for robust imbalanced classification, IEEE transactions on artificial intelligence 5 (1) (2023) 80–91

2023
[28]

Murtagh, Multilayer perceptrons for classification and regression, Neurocomputing 2 (5-6) (1991) 183–197

F. Murtagh, Multilayer perceptrons for classification and regression, Neurocomputing 2 (5-6) (1991) 183–197. 35

1991
[29]

Kruse, S

R. Kruse, S. Mostaghim, C. Borgelt, C. Braune, M. Steinbrecher, Multi- layer perceptrons, in: Computational intelligence: a methodological in- troduction, Springer, 2022, pp. 53–124

2022
[30]

Velayuthan, N

P. Velayuthan, N. Karunasingha, H. Nagahawaththa, B. G. Jayasekara, A. Hevapathige, Revisiting class imbalance: A generalized notion for oversampling, in: International Advanced Computing Conference, Springer, 2023, pp. 476–489

2023
[31]

Hornik, M

K. Hornik, M. Stinchcombe, H. White, Multilayer feedforward networks are universal approximators, Neural networks 2 (5) (1989) 359–366

1989
[32]

E. Jang, S. Gu, B. Poole, Categorical reparameterization with gumbel- softmax, stat 1050 (2017) 5

2017
[33]

Chahkoutahi, M

F. Chahkoutahi, M. Khashei, Influence of cost/loss functions on classifi- cation rate: A comparative study across diverse classifiers and domains, Engineering Applications of Artificial Intelligence 128 (2024) 107415

2024
[34]

Hornik, Approximation capabilities of multilayer feedforward net- works, Neural networks 4 (2) (1991) 251–257

K. Hornik, Approximation capabilities of multilayer feedforward net- works, Neural networks 4 (2) (1991) 251–257

1991
[35]

Y. S. Abu-Mostafa, The vapnik-chervonenkis dimension: Information versus complexity in learning, Neural Computation 1 (3) (1989) 312– 317

1989
[36]

J. W. Smith, J. E. Everhart, W. Dickson, W. C. Knowler, R. S. Jo- hannes, Using the adap learning algorithm to forecast the onset of dia- betes mellitus, in: Proceedings of the annual symposium on computer 36 application in medical care, American Medical Informatics Associatio...

1988
[37]

Fern´ andez, S

A. Fern´ andez, S. Garc ´ ıa, M. J. del Jesus, F. Herrera, A study of the be- haviour of linguistic fuzzy rule based classification systems in the frame- work of imbalanced data-sets, Fuzzy Sets and Systems 159 (18) (2008) 2378–2398

2008
[38]

Y. Dong, X. Wang, A new over-sampling approach: random-smote for learning from imbalanced data sets, in: Knowledge Science, Engineering and Management: 5th International Conference, KSEM 2011, Irvine, CA, USA, December 12-14, 2011. Proceedings 5, Springer, 2011, pp. 343–352

2011
[39]

W. N. Street, W. H. Wolberg, O. L. Mangasarian, Nuclear feature ex- traction for breast tumor diagnosis, in: Biomedical image processing and biomedical visualization, Vol. 1905, SPIE, 1993, pp. 861–870

1905
[40]

G. Pang, C. Shen, A. Van Den Hengel, Deep anomaly detection with deviation networks, in: Proceedings of the 25th ACM SIGKDD inter- national conference on knowledge discovery & data mining, 2019, pp. 353–362

2019
[41]

Menzies, J

T. Menzies, J. DiStefano, A. Orrego, R. Chapman, Assessing predictors of software defects, in: Proc. Workshop Predictive Software Models, 2004, pp. 1–11

2004
[42]

J. Read, B. Pfahringer, G. Holmes, E. Frank, Classifier chains for multi- label classification, Machine learning 85 (2011) 333–359. 37

2011
[43]

R. Das, A. Golatkar, S. P. Awate, Sparse kernel pca for outlier detection, in: 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA), IEEE, 2018, pp. 152–157

2018
[44]

N. W. S. Wardhani, M. Y. Rochayani, A. Iriany, A. D. Sulistyono, P. Lestantyo, Cross-validation metrics for evaluating classification per- formance on imbalanced data, in: 2019 international conference on com- puter, control, informatics and its applications (IC3INA), IEEE, 20...

2019
[45]

Kingma, Adam: a method for stochastic optimization, in: Int Conf Learn Represent, 2014

D. Kingma, Adam: a method for stochastic optimization, in: Int Conf Learn Represent, 2014

2014
[46]

Van Rossum, et al., Python programming language., in: USENIX annual technical conference, Vol

G. Van Rossum, et al., Python programming language., in: USENIX annual technical conference, Vol. 41, Santa Clara, CA, 2007, pp. 1–36

2007
[47]

Bisong, E

E. Bisong, E. Bisong, Introduction to scikit-learn, Building machine learning and deep learning models on google cloud platform: a compre- hensive guide for beginners (2019) 215–229

2019
[48]

Imambi, K

S. Imambi, K. B. Prakash, G. Kanagachidambaresan, Pytorch, Pro- gramming with TensorFlow: Solution for Edge Computing Applications (2021) 87–104

2021
[49]

McKinney, P

W. McKinney, P. Team, Pandas-powerful python data analysis toolkit, Pandas—Powerful Python Data Analysis Toolkit 1625
[50]

Bressert, Scipy and numpy: an overview for developers

E. Bressert, Scipy and numpy: an overview for developers. 38

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [1]

X. Guo, Y. Yin, C. Dong, G. Yang, G. Zhou, On the class imbalance problem, in: 2008 Fourth international conference on natural computa- tion, Vol. 4, IEEE, 2008, pp. 192–201

work page 2008

[2] [2]

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

work page 2019

[3] [3]

D. A. Cieslak, N. V. Chawla, A. Striegel, Combating imbalance in net- work intrusion datasets., in: GrC, 2006, pp. 732–737

work page 2006

[4] [4]

Japkowicz, S

N. Japkowicz, S. Stephen, The class imbalance problem: A systematic study, Intelligent data analysis 6 (5) (2002) 429–449

2002

[5] [5]

K. G. Al-Hashedi, P. Magalingam, Financial fraud detection applying data mining techniques: A comprehensive review from 2009 to 2019, Computer Science Review 40 (2021) 100402

work page 2021

[6] [6]

Khushi, K

M. Khushi, K. Shaukat, T. M. Alam, I. A. Hameed, S. Uddin, S. Luo, X. Yang, M. C. Reyes, A comparative performance analysis of data resampling methods on imbalance medical data, IEEE Access 9 (2021) 109960–109975

work page 2021

[7] [7]

A. I. Marqu´ es, V. Garc ´ ıa, J. S. S´ anchez, On the suitability of resampling techniques for the class imbalance problem in credit scoring, Journal of the Operational Research Society 64 (2013) 1060–1070

work page 2013

[8] [8]

N. V. Chawla, K. W. Bowyer, L. O. Hall, W. P. Kegelmeyer, Smote: 32 synthetic minority over-sampling technique, Journal of artificial intelli- gence research 16 (2002) 321–357

work page 2002