Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

RUBRIC: Realism--Utility Balanced Ranking for Imbalanced Classification

T0 review · 5 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read Selection, not generation, fixes oversampling: a realism-plus-utility filter provably tightens the generalization bound and lifts recall and F1 on fraud benchmarks.

desk verdict A sensible empirical post-filter for imbalanced learning, but the theory's central claim is assumed rather than proved, and duplicated table values put the empirical results in question. read the letter →

arxiv 2607.09816 v2 pith:KAIYYOY2 submitted 2026-07-10 cs.LG

Yanxuan Yu , Dong Liu , Eric Jiang , Shu Wang , Wenxiao Zhao , Jinxi Yu , Shaoyi Lu , Hui Pan
show 2 more authors
Renata Borovica-Gajic Ying Nian Wu
This is my paper · ORCID
classification cs.LG
keywords classimbalanceoversamplingsyntheticsamplefilteringmargin-basedgeneralizationboundrealism-utilitytrade-offsubmodularselectionfrauddetectiondiscriminatordensityratio
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

RUBRIC's central claim is that the fix for class-imbalanced learning is not a better oversampling generator but a filter that decides which synthetic candidates to keep. Each candidate is scored by a weighted combination of realism — the log-odds of a discriminator trained to distinguish real minority samples from synthetic ones — and boundary utility, a concave function of the candidate's margin under a simple classifier that rewards samples near, but not far beyond, the decision boundary. Under three regularity assumptions, the paper proves (Theorem 4.7) that selecting by this score monotonically tightens a margin-based excess-risk bound: realism filtering shrinks the distribution-shift term, and utility filtering raises a margin floor that exponentially suppresses contributions from near-boundary tail candidates. Empirically, on credit-card fraud and two other imbalanced tabular benchmarks, RUBRIC raises recall and F1 for most standard generators (SMOTE, ADASYN, Borderline-SMOTE) while holding ROC-AUC roughly constant, and a single λ dial trades F1/recall against AUPRC. If the paper is right, oversampling reduces to a quality-over-quantity selection problem: any generator can be upgraded by stacking a filter behind it.

What carries the argument

The load-bearing object is the realism–utility score s(˜x) = λ·g(margin_f(˜x)) + (1−λ)·log(D(˜x)/(1−D(˜x))): utility is a concave shaper (default g(t) = log(1+e^{t/τ})) giving diminishing returns to large margins, and realism is the discriminator log-odds, which for a Bayes-optimal discriminator equals the log density ratio log(p/q) between true minority density p and candidate density q. Ranking by s is assumed to induce a positive margin floor m0 (Assumption 4.3), converting the shaper's concavity into the exponential tail-suppression term C2 e^{−m0/τ}; the same discriminator feeds the plug-in KL estimate behind the total-variation bound δ̂ via Pinsker's inequality. Selection is monotone s

What would settle it

On the default Credit Card setup (SMOTE, logistic-regression boundary model, λ=0.5), record the selected set U and compute the realized margin distribution under f. If a non-trivial fraction of kept candidates has margin at or below zero — or if the realized m0 makes C2 e^{−m0/τ} negligible while recall/F1 gains persist — the margin-floor mechanism is not carrying the improvement. Companion check on a degraded configuration (SVM-SMOTE + RUBRIC, significantly negative ∆F1): did δ̂ fall and m0 rise there? If both moved but F1 fell, the bound does not govern threshold-dependent metrics. Also audi

Watch

Extended reading notes

Core claim

The paper's central result, Theorem 4.7 (Eq. 9), bounds the excess risk of ERM on the real data plus the selected synthetic set U: with probability at least 1−η, R(ĥ_w)−R(h*) ≤ O(LB/√N) + C1(δ̂+ε_est) + C2 e^{−m0/τ} + √(log(1/η)/N). δ̂ is a total-variation bound obtained from discriminator log-odds via Pinsker's inequality; ε_est is calibration error, additive so imperfection degrades the bound gracefully; m0 is a margin floor the selection threshold is assumed to impose on every kept candidate. Versus keeping all candidates, RUBRIC tightens the bound two ways at once — realism filtering shrinks δ̂ and utility filtering raises m0, driving the exponential term down — which is the stated mecha

Load-bearing premise

The load-bearing premise is Assumption 4.3: selection actually enforces a positive margin floor m0, so every kept candidate sits at least m0 from the decision boundary under f. The objective does not guarantee this — with small λ, realism dominates and low-margin candidates can be kept — and f is a linear model while the deployed classifier may be a tree ensemble, so f-margins need not be the margins that matter. If the floor fails, the C2 e^{−m0/τ} term in Eq. (9) has no bas

Editorial extensions

If this is right

  • Any oversampling generator becomes improvable: recall/F1 gains are reported with RUBRIC stacked on SMOTE, ADASYN, Borderline-SMOTE, and SMOTE-ENN, and even a generator-free 'NONE' candidate pool shows recall gains.
  • λ is a data-level frontier, not threshold tuning: small λ (≈0.3) preserves or improves AUPRC on extreme imbalance, large λ (≈0.7) buys recall and F1 at the cost of AUPRC, and users can move along this curve deliberately.
  • The theory identifies when filtering should fail — boundary-heavy generators combined with boundary-heavy selection amplify artifacts — and prescribes raising the realism weight or diversity strength in those cases.
  • Filtering is a one-time pre-training cost (≈12–85 s on the reported benchmarks) with no effect on inference latency, making it a practical drop-in for fraud and risk-screening pipelines.
  • The margin-based guarantee is cleanest when the auxiliary boundary model and the final classifier are both margin-based; the paper's own mixed Santander results with tree ensembles are consistent with weaker alignment there.

Reading between the lines

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

  • Editorial inference: because the realism score functions as a density-ratio estimate, the same ranking recipe should transfer to filtering synthetic pools beyond tabular data — image, text, or graph augmentations — wherever a discriminator and a task model can both be trained; the paper gestures at this but does not test it.
  • Editorial inference: the bound suggests a diagnostic the paper does not run — compute δ̂ and the empirical margin floor of the selected set per configuration. Where the bound tightens but F1 falls (SVM-SMOTE on Credit Card), the gap between margin-based theory and threshold-dependent metrics could be isolated as its own study.
  • Editorial inference: a natural ablation is to swap the linear boundary model f for the final classifier family (e.g., tree-based margin proxies); this should make Assumption 4.3 hold more often and would reveal how much of the gain depends on margin alignment.
  • Editorial inference: the facility-location diversity term is one coverage mechanism among several; replacing it with a determinantal point process would test whether diversity per se, rather than the realism–utility ranking, drives the recall gains.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. RUBRIC is a post-hoc filtering method for oversampling in imbalanced classification. It scores each synthetic candidate by a weighted combination of boundary utility u(˜x)=g(margin_f(˜x)) and discriminator log-odds realism r(˜x)=log(D/(1−D)), then selects K candidates maximizing Σ_{˜x∈U}(λu(˜x)+(1−λ)r(˜x))+γDiv(U). The paper claims that under Assumptions 4.1–4.3 this selection monotonically tightens a margin-based generalization bound (Theorem 4.7), and reports experiments on Credit Card Fraud, IEEE-CIS, and Santander showing improved recall/F1 for some generators with a λ-dependent trade-off. Code is released. The central theoretical claim, however, rests on a margin-floor assumption that the selection objective does not guarantee, and the appendix proof contains internal contradictions.

Significance. The core idea — selecting rather than generating synthetic samples, with an explicit realism–utility trade-off — is timely and the implementation is practical. The paper deserves credit for releasing code, reporting paired bootstrap confidence intervals, and openly discussing failure cases and λ sensitivity. If the margin-floor mechanism were derived from the algorithm or enforced by a hard constraint, the framework could be a useful contribution. As written, however, the advertised monotone-tightening result is not established: Assumption 4.3 is not implied by Eq. (3), Lemma 4.6 is vacuous under that assumption, and Appendix F does not provide a coherent proof of Theorem 4.7.

major comments (5)
  1. [§4, Assumption 4.3 and Eq. (3)] Assumption 4.3 states that RUBRIC selection induces a margin floor m0>0, but this is not a consequence of the score threshold s(˜x)≥t_K. Since s=λu+(1−λ)r and r is an unbounded logit, a candidate with very negative margin can still satisfy s≥t_K if r is sufficiently large; this is exactly the small-λ, realism-dominated regime. The utility shaper g is nondecreasing, but monotonicity of g alone gives no lower bound on margin from a lower bound on λu+(1−λ)r. The diversity term can also pull in low-margin points to improve facility-location coverage, contradicting Remark 4.4's claim that diversity 'can only raise this floor.' Consequently the C2 e^{−m0/τ} term in Eq. (9) is not justified by the algorithm; it is an additional assumption rather than a proved effect.
  2. [§4, Lemma 4.6] Under Assumption 4.3, every selected synthetic sample has margin ≥ m0, so the set of selected samples with margins below m0 is empty and the 'aggregate contribution' in Lemma 4.6 is identically zero. The displayed inequality max{0, g(0)−g(m0)} ≤ e^{−m0/τ} is a statement about the shaper at a single point, not a tail-suppression bound for the selected distribution; for m0>0 the left-hand side is 0. Thus Lemma 4.6 cannot provide the exponential suppression mechanism claimed in the interpretation after Theorem 4.7.
  3. [Appendix F (F.8, F.4, F.14)] Appendix F is not a coherent support for Theorem 4.7. Theorem F.8 states convergence to an ε-stationary point in O(1/ε²) iterations, but the proof concludes T=O(log(1/ε)); the PL constant is defined with µ=ε²/(2M²), so the linear-rate conclusion is invalid. Lemma F.7 applies McDiarmid with a bounded-difference coefficient Δ equal to the maximum gap between consecutive ordered scores; changing one score can change the sum of the top K by much more than that gap. Theorem F.14 claims O(m log m + K n d) yet the proof's own decomposition contains O(m n d), which is not bounded by the stated complexity. In addition, Definitions F.2–F.3 define RUBRIC-Score with the λ and (1−λ) weights interchanged relative to Eq. (3). These inconsistencies prevent verification of the theorem.
  4. [§4 vs. §6] Theorem 4.7 is a statement about ERM over linear predictors in an RKHS, with margins under the boundary model f (logistic regression or linear SVM). The experiments evaluate LightGBM, XGBoost, and RandomForest as final classifiers. Even if Assumption 4.3 held for f, it would not transfer to the margins of a tree ensemble, and no transfer or margin-preservation result is given. The theory therefore does not currently explain the reported F1/recall gains of the deployed classifiers.
  5. [Tables 4–6 and Tables 11–13] The abstract's claim that RUBRIC 'improves F1-macro and recall' is too strong in light of the paper's own paired-difference tables. Significant negative effects include Credit Card NONE ∆F1 = −0.208 [−0.364, −0.047], KMEANS-SMOTE ∆Recall = −0.052 [−0.074, −0.030], IEEE-CIS NONE ∆F1 = −0.210 [−0.369, −0.050], and Santander NONE ∆AUPRC = −0.162 [−0.248, −0.075]. The Discussion acknowledges some failure modes, but the abstract and conclusion should be qualified to reflect that gains are configuration-dependent.
minor comments (5)
  1. [Abstract and Introduction] The phrase 'monotonically tightens' is used as a central contribution, but no monotonicity theorem is stated. Remark 4.8 only says the bound decreases by decreasing δ̂ and increasing m0; neither monotonicity is proved.
  2. [§3.2 vs. Appendix F.2/F.3] The main text defines the selection score as s=λu+(1−λ)r, but Definition F.2 and F.3 write RUBRIC(˜x)=λ·R(˜x)+(1−λ)·U(˜x), swapping the roles of λ and 1−λ. This should be harmonized.
  3. [§4 vs. Definition F.4] Section 4 defines margin via the boundary model f as margin_f(x)=f(x), while Appendix Definition F.4 defines margin through nearest-neighbor distances. These are different objects and both are used without clarification.
  4. [Appendix F.14] The complexity proof's final equality drops the O(m n d) term from the utility computation. The stated complexity is only valid if the margin computation is done differently, which is not described.
  5. [Appendix F.16–F.24] Several appendix theorems (F.22–F.24) use Q-learning notation with states, actions, episodes, and H, S, A, but these are never connected to RUBRIC's selection problem. This material appears unrelated to the algorithm and should be removed or properly integrated.

Circularity Check

1 steps flagged · score 6.0 of 10

Tail-suppression term in Theorem 4.7 reduces to Assumption 4.3; monotone tightening is assumed, not derived.

  1. self definitional [Section 4, 'Score threshold and margin floor' paragraph; Assumption 4.3; Lemma 4.6; Theorem 4.7 Eq. (9); Remark 4.8]
    "Score-based selection induces a threshold t_K such that s(˜x) ≥ t_K for all ˜x ∈ U. By monotonicity of g, this implies a margin floor m0... Assumption 4.3: The selection rule induces a margin floor m0 > 0 such that margin_f(˜x) ≥ m0 for all ˜x ∈ U. Lemma 4.6: Suppose Assumption 4.3 holds... the aggregate contribution of synthetic samples with margins below m0 is bounded by C2 e^{−m0/τ}."

    Assumption 4.3 already asserts that every selected sample has margin_f ≥ m0, so U contains no samples with margin below m0. Lemma 4.6's 'aggregate contribution of synthetic samples with margins below m0' is therefore zero by definition, and the exponential term e^{−m0/τ} in Eq. (9) is simply the assumed floor relabeled as a suppression result. The advertised monotone tightening—Remark 4.8 claims RUBRIC works by 'increasing m0 via utility shaping'—is not derived from Eq. (3): since s = λu + (1−λ)r contains the unbounded realism logit r, a score threshold s ≥ t_K does not imply any positive margin floor, and no step shows filtering increases m0 relative to retaining all candidates. Thus the C2 e^{−m0/τ} term restates Assumption 4.3 rather than following from the selection algorithm.

full rationale

The empirical contribution is self-contained and externally benchmarked: paired bootstrap confidence intervals, negative results, ablations, and standard submodular-maximization guarantees are used without load-bearing self-citation or imported uniqueness theorems. The circularity is confined to the theoretical claim. Lemma 4.6's 'tail suppression' is a restatement of Assumption 4.3, and the claimed monotonic tightening by increasing m0 is not shown to be enforced by the realism–utility objective. Because the central theorem's exponential term reduces to the assumption by construction, the paper receives a partial-circularity score of 6 rather than 0–2.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

All free parameters are user-set; the theory does not predict their values. The two ad hoc assumptions—uniform discriminator calibration error ε_est and the margin floor m0—carry most of the theoretical conclusion, and neither is empirically verified.

free parameters (5)
  • λ (realism–utility trade-off) = 0.5 default; swept 0.0–1.0
    User-set knob; the paper's own λ-sweep shows it directly controls the AUPRC–Recall frontier, so any claimed improvement is conditional on this choice.
  • γ (diversity weight) = 0.1 default; ablation up to 0.3
    Sets strength of the submodular diversity term; the paper notes γ≥0.15 mitigates Borderline-SMOTE conflicts.
  • τ (utility shaper temperature) = 1.0 default
    Appears in the exponential tail bound e^{−m0/τ}; no principled selection criterion is given.
  • K (selection budget) = |n1| (match minority count)
    Ablations show K=0.5|n1| hurts recall and K=2|n1| gives <0.01 F1 gain; the budget choice changes conclusions.
  • Discriminator architecture/optimizer hyperparameters = hidden 128, dropout 0.2, lr 1e-3, batch 256 (Table 14)
    The realism score's quality depends on these settings; no sensitivity analysis is reported for the discriminator itself.
assumptions (6)
  • standard math Pinsker's inequality and Rademacher complexity bounds for margin classifiers are valid in the stated setting.
    Used in Lemma 4.2 and Theorem 4.7; these are standard tools.
  • standard math Greedy maximization of a monotone submodular function under a cardinality constraint achieves a (1−1/e) approximation.
    Prop. 3.1, citing Nemhauser et al. (1978) and Krause and Golovin (2014).
  • ad hoc to paper The discriminator logit satisfies |logit D(x) − log(p(x)/q(x))| ≤ ε_est uniformly over X (Assumption 4.1).
    No calibration experiment is reported; the TV-shift argument in Lemma 4.2 depends on this uniform error bound.
  • ad hoc to paper Selection via s = λu + (1−λ)r induces a margin floor m0 > 0 for all selected candidates (Assumption 4.3).
    This is the key premise of the tail-suppression term; it is not guaranteed when realism dominates the score or when f is not the final classifier.
  • domain assumption Margins computed by the boundary model f transfer to the final classifiers used in experiments (logistic regression, LightGBM, XGBoost, RandomForest).
    Section 4 restricts to linear RKHS margin classifiers, but Section 6 evaluates tree ensembles; no transfer argument is given.
  • domain assumption Minority density p and candidate density q are well-defined over R^d and synthetic candidates approximately live on the minority manifold M.
    Section 3.1 setup; needed for the density-ratio interpretation of the discriminator.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RUBRIC: Realism--Utility Balanced Ranking for Imbalanced Classification." pith.science (2026). https://pith.science/paper/KAIYYOY2

@misc{pith2026260709816,
  author       = {Pith},
  title        = {Pith review of: RUBRIC: Realism--Utility Balanced Ranking for Imbalanced Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KAIYYOY2}},
  note         = {Machine review of arXiv:2607.09816}
}
read the original abstract

Class imbalance poses a fundamental challenge in risk-sensitive applications such as fraud detection and medical diagnosis, where minority-class samples are scarce yet critical for accurate classification. Existing oversampling methods generate synthetic samples to rebalance class distributions; however, they often produce large numbers of low-quality candidates that distort decision boundaries or introduce artifacts, leading to overfitting and degraded generalization. In this work, we introduce RUBRIC, a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-over-quantity optimization problem. RUBRIC ranks candidates using a realism-utility trade-off: realism is quantified by a learned discriminator that distinguishes real samples from synthetic samples, while utility captures proximity to the decision boundary through a concave margin-based scoring function. We show that, under mild regularity conditions, the proposed filtering strategy monotonically tightens the generalization bound for margin-based classifiers by jointly reducing distribution shift and suppressing near-negative tail contributions. Through extensive experiments on credit-card fraud detection and other imbalanced benchmarks, we demonstrate that RUBRIC improves F1-macro and recall while maintaining comparable ROC-AUC across several generators. We also provide explicit lambda-sensitivity analysis to show how users can recover AUPRC when ranking quality is prioritized.

Figures

Figures reproduced from arXiv: 2607.09816 by the authors.

Figure 1
Figure 1. Performance improvement heatmap (mean across seeds; error bars in supplement) showing percentage [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. IEEE-CIS: average improvement summary with mean±std over seeds (error bars). This figure illustrates variability across runs and complements the heatmap aggregation. requiring high AUPRC should set λ ≤ 0.3; detection￾oriented settings benefit from λ ≥ 0.7. Runtime. RUBRIC filtering adds a one-time post￾processing cost before final classifier training ( [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 4
Figure 4. Performance comparison across three datasets [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figures from the paper (8 more)
Figure 5
Figure 5. Figure 5: Credit Card: SMOTE vs SMOTE+RUBRIC across metrics (mean over seeds; see main text for [PITH_FULL_IMAGE:figures/full_fig_p036_5.png]
Figure 6
Figure 6. Figure 6: IEEE-CIS: SMOTE vs SMOTE+RUBRIC across metrics (mean over seeds; see main text for protocol). [PITH_FULL_IMAGE:figures/full_fig_p037_6.png]
Figure 7
Figure 7. Figure 7: Santander: SMOTE vs SMOTE+RUBRIC across metrics (mean over seeds; see main text for protocol). [PITH_FULL_IMAGE:figures/full_fig_p038_7.png]
Figure 8
Figure 8. Figure 8: Credit Card improvement heatmap (mean across seeds). [PITH_FULL_IMAGE:figures/full_fig_p038_8.png]
Figure 9
Figure 9. Figure 9: Santander improvement heatmap (mean across seeds). [PITH_FULL_IMAGE:figures/full_fig_p039_9.png]
Figure 10
Figure 10. Figure 10: Credit Card: average improvement summary (mean [PITH_FULL_IMAGE:figures/full_fig_p039_10.png]
Figure 11
Figure 11. Figure 11: IEEE-CIS: average improvement summary (mean [PITH_FULL_IMAGE:figures/full_fig_p040_11.png]
Figure 12
Figure 12. Figure 12: Santander: average improvement summary (mean [PITH_FULL_IMAGE:figures/full_fig_p040_12.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Se-DPO: Self-Evolving Token Credit for Direct Preference Optimization

    cs.CL 2026-08 reject novelty 5.0 of 10

    Se-DPO assigns evolving per-token credits from implicit reward magnitude and reference entropy during DPO training, improving instruction-following win rates but with an internally inconsistent derivation.

Reference graph

Works this paper leans on

61 extracted references · 1 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Proceedings of the International Conference on Knowledge Discovery and Data Mining (KDD) , pages=

    SMOTE: Synthetic Minority Over-sampling Technique , author=. Proceedings of the International Conference on Knowledge Discovery and Data Mining (KDD) , pages=

  2. [2]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Learning from imbalanced data , author=. IEEE Transactions on Knowledge and Data Engineering , volume=

  3. [3]

    International Conference on Intelligent Computing , pages=

    Borderline-SMOTE: A New Over-Sampling Method in Imbalanced Data Sets Learning , author=. International Conference on Intelligent Computing , pages=. 2005 , organization=

    2005

  4. [4]

    IEEE International Joint Conference on Neural Networks (IJCNN) , pages=

    ADASYN: Adaptive synthetic sampling approach for imbalanced learning , author=. IEEE International Joint Conference on Neural Networks (IJCNN) , pages=

  5. [5]

    International Journal of Knowledge Engineering and Soft Data Paradigms , year=

    Borderline over-sampling for imbalanced data classification , author=. International Journal of Knowledge Engineering and Soft Data Paradigms , year=

  6. [6]

    EPIA Conference on Artificial Intelligence , pages=

    SMOTE for Regression with K-Means and Kernel Functions , author=. EPIA Conference on Artificial Intelligence , pages=

  7. [7]

    Progress in Artificial Intelligence , volume=

    Learning from imbalanced data: open challenges and future directions , author=. Progress in Artificial Intelligence , volume=

  8. [8]

    Expert Systems with Applications , volume=

    Effective data generation for imbalanced learning using conditional generative adversarial networks , author=. Expert Systems with Applications , volume=

Show all 61 references
  1. [9]

    Information Sciences , volume=

    Using generative adversarial networks for improving classification effectiveness in credit card fraud detection , author=. Information Sciences , volume=

  2. [10]

    Expert Systems with Applications , volume=

    Effective Data Generation for Imbalanced Learning Using Conditional GANs , author=. Expert Systems with Applications , volume=. 2018 , publisher=

    2018

  3. [11]

    PLoS ONE , volume=

    The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets , author=. PLoS ONE , volume=

  4. [12]

    ICML , pages=

    The Relationship Between Precision-Recall and ROC Curves , author=. ICML , pages=

  5. [13]

    AISTATS , year=

    Beta Calibration: a Well-Founded and Easily Implemented Improvement on Logistic Calibration for Binary Classifiers , author=. AISTATS , year=

  6. [14]

    2018 , publisher=

    Foundations of Machine Learning , author=. 2018 , publisher=

    2018

  7. [15]

    2018 , url=

    Credit Card Fraud Detection , author=. 2018 , url=

    2018

  8. [16]

    Information Sciences , volume=

    Oversampling for Imbalanced Learning Based on K-Means and SMOTE , author=. Information Sciences , volume=. 2018 , publisher=

    2018

  9. [17]

    Journal of Machine Learning Research , volume=

    Rademacher and Gaussian Complexities: Risk Bounds and Structural Results , author=. Journal of Machine Learning Research , volume=

  10. [18]

    The Annals of Statistics , volume =

    Learning Models with Uniform Performance via Distributionally Robust Optimization , author =. The Annals of Statistics , volume =. 2021 , publisher =. doi:10.1214/20-AOS2004 , url =. 1810.08750 , archivePrefix =

    2021 arXiv

  11. [19]

    2015 IEEE Symposium Series on Computational Intelligence (SSCI) , pages=

    Calibrating Probability with Undersampling for Unbalanced Classification , author=. 2015 IEEE Symposium Series on Computational Intelligence (SSCI) , pages=. 2015 , organization=

    2015

  12. [20]

    2019 , url=

    Santander Customer Transaction Prediction , author=. 2019 , url=

    2019

  13. [21]

    Pacific-Asia Conference on Knowledge Discovery and Data Mining , pages=

    Safe-Level-SMOTE: Safe-Level-Synthetic Minority Over-sampling Technique for Handling the Class Imbalanced Problem , author=. Pacific-Asia Conference on Knowledge Discovery and Data Mining , pages=. 2009 , organization=

    2009

  14. [22]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    MWMOTE--Majority Weighted Minority Oversampling Technique for Imbalanced Data Set Learning , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2014 , publisher=

    2014

  15. [23]

    Journal of Artificial Intelligence Research , volume=

    SMOTE for learning from imbalanced data: progress and challenges, marking the 15-year anniversary , author=. Journal of Artificial Intelligence Research , volume=

  16. [24]

    2018 , note=

    BAGAN: Data Augmentation with Balancing GAN , author=. 2018 , note=

    2018

  17. [25]

    IEEE Access , volume=

    Handling Imbalanced Data in GAN-based Image Generation via Adaptive Weighted Discriminator , author=. IEEE Access , volume=. 2020 , publisher=

    2020

  18. [26]

    Advances in Neural Information Processing Systems , volume=

    Generative Adversarial Nets , author=. Advances in Neural Information Processing Systems , volume=

  19. [27]

    2014 , note=

    Conditional Generative Adversarial Nets , author=. 2014 , note=

    2014

  20. [28]

    2015 , note=

    Explaining and Harnessing Adversarial Examples , author=. 2015 , note=

    2015

  21. [29]

    2012 , publisher=

    Density Ratio Estimation in Machine Learning , author=. 2012 , publisher=

    2012

  22. [30]

    Annals of Statistics , volume=

    Empirical margin distributions and bounding the generalization error of combined classifiers , author=. Annals of Statistics , volume=. 2002 , publisher=

    2002

  23. [31]

    IEEE Transactions on Information Theory , volume=

    The sample complexity of pattern classification with neural networks: the size of the weights is more important than the size of the network , author=. IEEE Transactions on Information Theory , volume=. 1998 , publisher=

    1998

  24. [32]

    Annals of Statistics , volume=

    Boosting the margin: A new explanation for the effectiveness of voting methods , author=. Annals of Statistics , volume=. 1998 , publisher=

    1998

  25. [33]

    2009 , publisher=

    Introduction to Nonparametric Estimation , author=. 2009 , publisher=

    2009

  26. [34]

    1964 , publisher=

    Information and Information Stability of Random Variables and Processes , author=. 1964 , publisher=

    1964

  27. [35]

    2012 , publisher=

    Elements of Information Theory , author=. 2012 , publisher=

    2012

  28. [36]

    Proceedings of the 17th International Joint Conference on Artificial Intelligence (IJCAI) , pages=

    The Foundations of Cost-Sensitive Learning , author=. Proceedings of the 17th International Joint Conference on Artificial Intelligence (IJCAI) , pages=

  29. [37]

    IEEE Transactions on Knowledge and Data Engineering , volume =

    Training Cost-Sensitive Neural Networks with Methods Addressing the Class Imbalance Problem , author =. IEEE Transactions on Knowledge and Data Engineering , volume =. 2006 , month = jan, doi =

    2006

  30. [38]

    IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) , volume=

    A review on ensembles for the class imbalance problem: bagging-, boosting-, and hybrid-based approaches , author=. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) , volume=. 2012 , publisher=

    2012

  31. [39]

    IEEE Transactions on Systems, Man, and Cybernetics---Part A: Systems and Humans , volume =

    RUSBoost: A Hybrid Approach to Alleviating Class Imbalance , author =. IEEE Transactions on Systems, Man, and Cybernetics---Part A: Systems and Humans , volume =. 2010 , doi =

    2010

  32. [40]

    Expert Systems with Applications , volume=

    Learning from class-imbalanced data: Review of methods and applications , author=. Expert Systems with Applications , volume=. 2017 , publisher=

    2017

  33. [41]

    Journal of Big Data , volume=

    Survey on deep learning with class imbalance , author=. Journal of Big Data , volume=. 2019 , publisher=

    2019

  34. [42]

    Imbalanced Learning: Foundations, Algorithms, and Applications , pages=

    Foundations of Imbalanced Learning , author=. Imbalanced Learning: Foundations, Algorithms, and Applications , pages=. 2013 , publisher=

    2013

  35. [43]

    ACM Computing Surveys (CSUR) , volume=

    A survey of predictive modeling on imbalanced domains , author=. ACM Computing Surveys (CSUR) , volume=. 2017 , publisher=

    2017

  36. [44]

    Advances in Large Margin Classifiers , pages=

    Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods , author=. Advances in Large Margin Classifiers , pages=. 1999 , publisher=

    1999

  37. [45]

    Proceedings of the 34th International Conference on Machine Learning (ICML) , pages=

    On Calibration of Modern Neural Networks , author=. Proceedings of the 34th International Conference on Machine Learning (ICML) , pages=. 2017 , organization=

    2017

  38. [46]

    Proceedings of the IEEE International Conference on Computer Vision , pages=

    Focal Loss for Dense Object Detection , author=. Proceedings of the IEEE International Conference on Computer Vision , pages=

  39. [47]

    Advances in Neural Information Processing Systems , volume=

    On the relationship between binary classification, bipartite ranking, and binary class probability estimation , author=. Advances in Neural Information Processing Systems , volume=

  40. [48]

    Machine Learning , volume=

    A theory of learning from different domains , author=. Machine Learning , volume=. 2010 , publisher=

    2010

  41. [49]

    Advances in Neural Information Processing Systems , volume=

    Domain Adaptation with Multiple Sources , author=. Advances in Neural Information Processing Systems , volume=

  42. [50]

    2002 , publisher=

    Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond , author=. 2002 , publisher=

    2002

  43. [51]

    Journal of Machine Learning Research , volume=

    Stability and generalization , author=. Journal of Machine Learning Research , volume=

  44. [52]

    2014 , publisher=

    Understanding Machine Learning: From Theory to Algorithms , author=. 2014 , publisher=

    2014

  45. [53]

    2025 , eprint =

    Boundary-Aware Adversarial Filtering for Reliable Diagnosis under Extreme Class Imbalance , author =. 2025 , eprint =

    2025

  46. [54]

    Machine Learning , volume =

    A survey on semi-supervised learning , author =. Machine Learning , volume =. 2020 , doi =

    2020

  47. [55]

    Mathematical Programming , volume =

    An analysis of approximations for maximizing submodular set functions---I , author =. Mathematical Programming , volume =. 1978 , doi =

    1978

  48. [56]

    Tractability: Practical Approaches to Hard Problems , editor =

    Submodular Function Maximization , author =. Tractability: Practical Approaches to Hard Problems , editor =

  49. [57]

    2025 , address =

    Liu, Dong and Yu, Yanxuan , booktitle =. 2025 , address =

    2025

  50. [58]

    2026 , address =

    Liu, Dong and Yu, Yanxuan , booktitle =. 2026 , address =

    2026

  51. [59]

    2025 , publisher =

    Liu, Dong and Yu, Yanxuan , booktitle =. 2025 , publisher =

    2025

  52. [60]

    Liu, Dong and Yu, Yanxuan and Lengerich, Ben and Wu, Ying Nian , booktitle =

  53. [61]

    2026 , address =

    Liu, Dong and Yu, Yanxuan , booktitle =. 2026 , address =. doi:10.1145/3748173.3779188 , url =

    2026

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.