Pith. sign in

REVIEW 4 major objections 5 minor 41 references

Metric-DST: Mitigating Selection Bias Through Diversity-Guided Semi-Supervised Metric Learning

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

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

Pith's one-line read Metric-DST claims that selecting diverse pseudo-labeled samples via random points in a metric-learned embedding space counters selection bias better than conventional confidence-based self-training.

desk verdict Metric-DST is a plausible, carefully evaluated diversity-guided self-training method that mostly holds its own under selection bias, but the core selection step is underspecified and DCAST is never compared. read the letter →

arxiv 2411.18442 v2 pith:27INDJCR submitted 2024-11-27 cs.LG cs.AI

Yasin I. Tepeli , Mathijs de Wolf , Joana P. Gonçalves This is my paper · ORCID
classification cs.LGcs.AI
keywords selectionbiasmetriclearningself-trainingdiversitysemi-supervisedfairnesspseudo-labelingsyntheticlethality
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 proposes Metric-DST, a self-training strategy that mitigates selection bias by selecting a diverse set of pseudo-labeled samples instead of the most confident ones. It claims that by learning a metric embedding from labeled data and picking pseudo-labeled samples near random points in that embedding space, the method counters the confirmation bias that makes conventional self-training reinforce existing data bias. On generated and real-world datasets with induced bias, and on synthetic lethality prediction with intrinsic bias, Metric-DST mostly preserved or modestly improved AUROC/AUPRC compared to supervised learning, while conventional self-training (Metric-ST) often degraded performance. The paper argues this makes diversity-guided selection a flexible, widely applicable way to improve model robustness under selection bias.

What carries the argument

The load-bearing object is the class-contrastive metric embedding f_θ: X → Z, trained with a contrastive loss (Eq. 1) that pulls same-class pairs within a positive margin and pushes different-class pairs beyond a negative margin. This embedding serves two roles: it gives a distance-based confidence for pseudo-labeling via weighted kNN (Eq. 2), and it provides the geometry in which diversity is sampled, by generating uniform random points in the bounded embedding space and selecting the nearest pseudo-labeled sample above the relaxed threshold μ. Diversity is further enforced by selecting equal numbers per class. The selection mechanism is designed to replace the highest-confidence rule of conventional self-training, which the paper argues reinforces bias.

What would settle it

If the unlabeled data are drawn from exactly the same biased distribution as the labeled set, Metric-DST should provide no benefit over supervised learning; an experiment that constructs such an identically biased unlabeled pool and shows AUROC does not improve would falsify the core mechanism. A second check: if the embedding distances are not bounded (e.g., the contrastive loss without output normalization permits d > 1), the (1 − d) confidence weights in Eq. 2 become negative, and the method's pseudo-label confidence is ill-defined.

Watch

Extended reading notes

Core claim

The central claim is that diversity-guided selection of pseudo-labeled samples, mediated by a metric-learned embedding, produces models more robust to selection bias than models trained on biased labeled data alone or by confidence-based self-training. In the proposed scheme, each self-training iteration learns a contrastive metric transformation from labeled samples, uses weighted kNN in the embedding to pseudo-label unlabeled samples, and then selects p/2 positive and p/2 negative samples by generating random coordinates in the embedding space and accepting the nearest pseudo-labeled sample whose confidence exceeds a relaxed threshold μ. The paper reports that this procedure yields no significant performance losses relative to supervised learning across most tested settings and a small number of significant gains, whereas the non-diverse Metric-ST baseline sometimes significantly underperforms supervised learning. The evidence includes a median AUROC improvement on the moons dataset with 100-sample delta bias, significant gains on Fire, Raisin, and Adult with 60 selections, and a significant AUPRC gain on CESC in synthetic lethality prediction under randomized and double-holdout splits.

Load-bearing premise

The method assumes that the unlabeled data actually cover the underrepresented regions of the population and that the metric embedding learned from the biased labeled set is a faithful geometry for diversity sampling, so that random points in that space can find genuinely diverse samples.

Editorial extensions

If this is right

  • Metric-DST preserves or improves AUROC/AUPRC relative to biased supervised learning across generated, real-world, and intrinsic-bias tasks, with a small number of significant gains.
  • Conventional self-training (Metric-ST) can significantly degrade performance under selection bias, whereas Metric-DST does not.
  • The diversity strategy selects a more heterogeneous set of pseudo-labeled samples, as evidenced by larger average pairwise distances and fewer samples drawn from a single dominant cluster in the BRCA synthetic lethality analysis.
  • Because the framework is built on a metric embedding plus kNN, it can be combined with different classifier architectures, making it a flexible component for bias mitigation.

Reading between the lines

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

  • The method's success is contingent on unlabeled coverage; if the unlabeled pool inherits the same bias as the labeled set, diversity sampling cannot recover the missing regions, so the method would reduce to supervised learning with extra noise. This is an implicit boundary condition the paper does not test.
  • The random-point selection could be replaced by deterministic coverage strategies such as farthest-point sampling or K-center selection, which might be more reproducible and sample-efficient; comparing these would isolate the contribution of randomness.
  • The paper evaluates robustness via AUROC/AUPRC but not group-fairness metrics such as demographic parity or equalized odds; testing whether diversity-guided selection improves fairness on protected subgroups would connect the method directly to the fairness framing.
  • The bounded-distance assumption behind the (1 − d) confidence weights is not verified; enforcing an explicit unit-ball regularization on the embedding would make the confidence computation principled and likely more stable across datasets.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes Metric-DST, a self-training framework that combines metric learning with diversity-guided pseudo-label selection. At each iteration, a contrastively trained embedding network maps labeled and unlabeled samples to a low-dimensional latent space; unlabeled samples are pseudo-labeled by weighted kNN; and p/2 samples per class are selected by sampling random coordinates in [0,1]^d and taking the nearest pseudo-labeled sample whose confidence exceeds a relaxed threshold. The authors compare Metric-DST against Metric-ST (confidence-based self-training) and supervised metric learning on generated moons and higher-dimensional datasets with induced bias, eight real-world benchmark datasets with induced hierarchy bias, and synthetic lethality prediction with intrinsic bias, using shared folds and paired Wilcoxon tests. The main finding is that Metric-DST rarely hurts and occasionally improves AUROC/AUPRC relative to supervised learning, while Metric-ST sometimes degrades performance; a qualitative analysis on BRCA shows that Metric-DST selects samples from more diverse clusters than Metric-ST.

Significance. If the method is made fully specified, the paper's contribution is useful: it addresses a real problem (confirmation bias in self-training), proposes a simple diversity mechanism that is compatible with a metric-learning backbone, and evaluates it carefully with the same folds, controls that separate bias effects from sample-size effects, and paired significance tests. The released code and data are a concrete strength. However, the empirical gains are modest and only a small number of comparisons reach significance; the strongest support is the consistent absence of significant losses rather than consistent gains. The central diversity mechanism additionally relies on a bounded or normalized embedding that is not established in the manuscript, so the method as written is underspecified. With revision, the core idea is defensible, but the current text overstates the universality and robustness of the approach.

major comments (4)
  1. [Experimental procedures, 'Selection of diverse pseudo-labeled samples' and Eq. (2)] The diversity-selection step is underspecified at exactly the point that distinguishes Metric-DST from Metric-ST. Eq. (2) treats (1 - d_{i,n}) as a confidence weight, which is only valid if all kNN distances lie in [0,1]; otherwise \bar{y}_i can fall outside [0,1] and the relaxed threshold \mu has no probabilistic meaning. The contrastive loss in Eq. (1) does not enforce this bound: once a negative pair exceeds m_neg it no longer contributes to the loss, so distances between different-class samples can grow without penalty, and no normalization of z or d is stated. Likewise, random candidate points are drawn as coordinates in [0,1], but the paper never establishes that the learned embeddings f_\theta(x) are confined to [0,1]^d; if they are not, the 'closest pseudo-labeled sample' to a random point is a direction-dependent proxy rather than a principled diversity sample. The paper's own limitation note in the Conclusion, that unpopulated regions of the embedding may be sampled, reinforces this concern. The released code may contain an implicit normalization, but the manuscript must state it, justify it, and report its effect.
  2. [Results, 'Higher-dimensional two-cluster datasets and hierarchy bias'] The conclusion that Metric-DST 'preserve[d] the performance obtained with supervised learning or deliver[ed] more robust models in all contexts' is contradicted by the paper's own result for the 200-sample selection on the 64-dimensional dataset with 80% informative features, where Metric-DST was significantly worse than supervised learning (p = 0.004). This exception should be acknowledged wherever the no-harm claim is made, and the authors should discuss why the method failed in that setting.
  3. [Results, real-world benchmark datasets and synthetic lethality prediction] The evidence for improved bias mitigation rests on a small number of nominally significant Wilcoxon tests selected from a much larger set of comparisons: Fig. 2d reports a few significant gains among eight datasets at two selection sizes, and Fig. 3 reports isolated gains in CESC and one LUAD split. Without correction for multiple testing, the observed significant differences could partly reflect chance. Please report adjusted p-values (e.g., FDR) or clearly present the analysis as exploratory, and base the central claim on the consistent absence of harm rather than on isolated significant gains.
  4. [Results, 'Metric-DST promotes diversity in selected pseudo-labeled samples' and Fig. 3d] The diversity verification in Fig. 3d measures average Euclidean distances between selected pseudo-labeled samples in the same learned embedding space in which the random-point selection operates, so larger distances are partly a consequence of the selection mechanism rather than an independent confirmation of diversity. The CDH1-cluster analysis in Fig. 4 is more convincing because it is expressed in original-feature and biological terms. Please either report diversity in a space not used for selection, or explicitly label Fig. 3d as a sanity check of the mechanism and present the feature-space analysis as the diversity evidence.
minor comments (5)
  1. [Conclusion and Fig. 4 text] 'Utimately' is a typo for 'Ultimately', and the paragraph around Fig. 4 refers to the 'CHD1' cluster after spelling the gene as 'CDH1' in the same discussion.
  2. [Results, real-world benchmark datasets] 'Event though' should be 'Even though' in the first sentence of the real-world benchmark section.
  3. [Experimental procedures, Evaluation of Metric-DST] The values of the contrastive margins m_pos and m_neg in Eq. (1) are not reported; please include them for reproducibility.
  4. [Experimental procedures, Synthetic lethality prediction] For the synthetic lethality experiments, \mu and p were selected per run by grid search on validation loss; the manuscript should state whether the same per-run tuning budget was applied to Metric-ST and supervised baselines, so that the comparison does not favor Metric-DST through a larger hyperparameter search.
  5. [Introduction and Conclusion] The claim that Metric-DST 'can be used with virtually any type of classifier' is too strong given that both pseudo-labeling and diversity selection are defined through the metric embedding and kNN; please qualify the claim.

Circularity Check

1 steps flagged · score 3.0 of 10

Central bias-mitigation evaluation is external and not circular; only the diversity-verification check is self-referential because it measures the same embedding distances that the selection procedure actively maximizes.

  1. self definitional [Results, subsection 'Metric-DST promotes diversity in selected pseudo-labeled samples' (Fig. 3d); Methods, 'Selection of diverse pseudo-labeled samples']
    "To verify if the diversity approach of Metric-DST was able to select more diverse samples, we analyzed the Euclidean pairwise distances between pseudo-labeled samples assigned to the same class label in the learned embedding space... The distances were larger on average for pseudo-samples selected by Metric-DST, confirming a more heterogeneous sample selection compared to Metric-ST."

    The selection procedure defines diversity in exactly the same terms that the verification measures: a random candidate point is generated in the learned embedding space and 'the pseudo-labeled sample closest to the candidate point is identified based on the Euclidean distance'. Therefore the Fig. 3d check that Metric-DST-selected samples have larger Euclidean pairwise distances in that same embedding is not independent evidence of diversity; it largely restates the selection objective. The claim that 'Metric-DST promotes diversity' is thus partly self-definitional. This does not by itself circularize the main AUROC/AUPRC evaluation, which is computed on held-out test folds rather than on the embedding distances used for selection.

full rationale

The core performance claim is benchmarked externally: each trained model is evaluated on held-out test folds for generated and real-world data, and on test splits for synthetic lethality, so the main result is not equivalent to the method's inputs. Citations to the authors' prior work (DCAST for hierarchy bias, ELISL for SL features and experimental setup) supply datasets and protocols, not the load-bearing argument that diversity selection improves bias robustness; those evaluations are independent of the citations. The only notable self-referential element is the Fig. 3d diversity verification, which measures Euclidean pairwise distances in the same learned embedding space in which Metric-DST's random-point selection explicitly maximizes spread; that check is an internal consistency demonstration rather than a separate confirmation. The unspecified normalization of distances in Eq. 2 (confidence weights require distances in [0,1] for a convex combination, and random coordinates in [0,1]^d require bounded embeddings) is a specification/correctness risk, not a circularity, because it does not make the test-set results reduce to the method's assumptions. Overall, the central claim retains independent empirical content; the circularity score is low.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central empirical claims rest on a small set of hyperparameters (mu, p, k, margins, architecture) that are either set by hand, by a heuristic (p=floor(sqrt(n))), or by validation-based grid search. No parameter is fitted to the test-set results themselves, so the ledger is unremarkable for an applied ML paper.

free parameters (5)
  • confidence threshold mu = 0.9 (CV); grid-searched 0.70-0.95 (SL)
    Controls eligibility of pseudo-labeled samples for inclusion. Central to the diversity selection procedure; not derived from theory.
  • pseudo-labeled sample count p = floor(sqrt(n)) (CV); grid-searched 4-50 (SL)
    Selects p/2 samples per class each iteration. Determines how quickly unlabeled data are absorbed and how much pseudo-label noise enters.
  • kNN neighbors k = 5
    Used for pseudo-labeling and confidence estimation; a design choice.
  • contrastive margins m_pos, m_neg = not specified
    Shape the embedding geometry that diversity sampling depends on; no values given.
  • architecture and batch size = hidden 8, output 2, batch 64
    Fixed neural network configuration; output dimension 2 is where random candidate points are drawn.
assumptions (4)
  • domain assumption Unlabeled data are representative of the underlying population
    Introduction states that unlabeled data may offer insight into the true underlying distribution; in experiments bias is induced only on the labeled subset, so the unlabeled set is unbiased by construction.
  • domain assumption The metric embedding learned from biased labeled data captures class structure of the full population well enough for diversity sampling
    Methods section 'Learning of a metric embedding function' learns f_theta from labeled samples only; if the labeled set is severely biased, the embedding may be distorted.
  • domain assumption Pseudo-labels above the relaxed confidence threshold are sufficiently accurate to benefit training
    Selection of diverse pseudo-labeled samples requires confidence to surpass threshold mu, but the paper does not analyze the relationship between confidence and pseudo-label accuracy.
  • ad hoc to paper Embedding space is bounded and distances are normalized to [0,1] so that (1 - d) is a valid weight
    Eq. 2 uses (1 - di,n) as a similarity weight; the paper calls the space bounded but never specifies the normalization, so this assumption is unverified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Metric-DST: Mitigating Selection Bias Through Diversity-Guided Semi-Supervised Metric Learning." pith.science (2026). https://pith.science/paper/27INDJCR

@misc{pith2026241118442,
  author       = {Pith},
  title        = {Pith review of: Metric-DST: Mitigating Selection Bias Through Diversity-Guided Semi-Supervised Metric Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/27INDJCR}},
  note         = {Machine review of arXiv:2411.18442}
}
read the original abstract

Selection bias poses a critical challenge for fairness in machine learning, as models trained on data that is less representative of the population might exhibit undesirable behavior for underrepresented profiles. Semi-supervised learning strategies like self-training can mitigate selection bias by incorporating unlabeled data into model training to gain further insight into the distribution of the population. However, conventional self-training seeks to include high-confidence data samples, which may reinforce existing model bias and compromise effectiveness. We propose Metric-DST, a diversity-guided self-training strategy that leverages metric learning and its implicit embedding space to counter confidence-based bias through the inclusion of more diverse samples. Metric-DST learned more robust models in the presence of selection bias for generated and real-world datasets with induced bias, as well as a molecular biology prediction task with intrinsic bias. The Metric-DST learning strategy offers a flexible and widely applicable solution to mitigate selection bias and enhance fairness of machine learning models.

Figures

Figures reproduced from arXiv: 2411.18442 by the authors.

Figure 1
Figure 1. Overview of the Metric-DST methodology. A Metric-DST iteration encompasses 1) training a metric learning model on labeled data that can be used to transform both labeled and unlabeled samples into an embedding space, 2) obtaining predicted pseudo-labels and model confidence values for unlabeled samples using k-nearest neighbors (kNN) on the embed￾ding space representations, 3) selecting diverse pseudo-labeled sample… view at source ↗
Figure 2
Figure 2. Mitigation of selection bias induced to generated and real-world benchmark data. (a) Samples selected by delta bias (∆0 = ∆1 = (0, 0) for classes 0 and 1) highlighted on a scatter plot of the artificially generated 2D moons dataset. Performance (AUROC) of supervised and semi-supervised Metric-(D)ST methods using metric learning and kNN on: (b) generated 2D moons dataset of 2000 samples with four delta bias induction… view at source ↗
Figure 3
Figure 3. Mitigation of intrinsic selection bias for synthetic lethality prediction. Prediction performance (AUPRC) of synthetic lethality prediction models trained and tested per cancer type using supervised learning or the semi-supervised Metric-ST and Metric-DST methods for 10 train/test splits. Three types of splits were used to control the degree of similarity in selection bias between the train and test sets: (a) Random… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: UMAP projections of the SL dataset for BRCA. On the left, the training samples are highlighted before the training. The top right plot shows the pseudo-labeled samples selected by Metric-ST and the pseudo-labeled samples selected by Metric-ST during the training. The n…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages

  1. [1]

    Correcting sample selection bias for image classifica- tion

    D. Wu, D. Lin, L. Y ao, and W. Zhang. “Correcting sample selection bias for image classifica- tion”. In: 3rd International Conference on Intelligent System and Knowledge Engineering . Vol. 1. 2008, pp. 1214–1220

    work page 2008

  2. [2]

    Active and Semisupervised Learning for the Classification of Remote Sensing Images

    C. Persello and L. Bruzzone. “Active and Semisupervised Learning for the Classification of Remote Sensing Images”. In: IEEE Transactions on Geoscience and Remote Sensing 52.11 (2014), pp. 6937–6956

    work page 2014

  3. [3]

    Active learning to overcome sample selection bias: application to photometric variable star classification

    J. W. Richards, D. L. Starr, H. Brink, A. A. Miller, J. S. Bloom, N. R. Butler, J. B. James, J. P . Long, and J. Rice. “Active learning to overcome sample selection bias: application to photometric variable star classification”. In: The Astrophysical Journal 744.2 (Dec. 2011), p. 192

    work page 2011

  4. [4]

    Reject inference in credit scoring using a three-way decision and safe semi-supervised support vector machine

    F . Shen, Z. Y ang, X. Zhao, and D. Lan. “Reject inference in credit scoring using a three-way decision and safe semi-supervised support vector machine”. In: Information Sciences 606 (Aug. 2022), pp. 614–627

    work page 2022

  5. [5]

    Impact of Query Sample Selection Bias on Information Retrieval System Rank- ing

    M. Melucci. “Impact of Query Sample Selection Bias on Information Retrieval System Rank- ing”. In: IEEE International Conference on Data Science and Advanced Analytics (DSAA) . 2016, pp. 341–350

    work page 2016

  6. [6]

    Domain Adaptation with Structural Correspondence Learning

    J. Blitzer, R. McDonald, and F . Pereira. “Domain Adaptation with Structural Correspondence Learning”. In: Proceedings of the 2006 Conference on Empirical Methods in Natural Lan- guage Processing. EMNLP ’06. Sydney, Australia: Association for Computational Linguis- tics, 2006, pp. 120–128. ISBN : 1932432736

    work page 2006

  7. [7]

    Unsupervised Visual Domain Adaptation Using Subspace Alignment

    B. Fernando, A. Habrard, M. Sebban, and T. Tuytelaars. “Unsupervised Visual Domain Adaptation Using Subspace Alignment”. In: IEEE International Conference on Computer Vision. 2013, pp. 2960–2967

    work page 2013

  8. [8]

    Feature-Level Domain Adaptation

    W. M. Kouw, L. J. P . Van Der Maaten, J. H. Krijthe, and M. Loog. “Feature-Level Domain Adaptation”. In: Journal of Machine Learning Research 17.1 (Jan. 2016), pp. 5943–5974. ISSN : 1532-4435

    work page 2016

Show all 41 references
  1. [9]

    Learning and Evaluating Classifiers under Sample Selection Bias

    B. Zadrozny. “Learning and Evaluating Classifiers under Sample Selection Bias”. In: Proceedings of the Twenty-First International Conference on Machine Learning . ICML ’04. Banff, Alberta, Canada: Association for Computing Machinery, 2004, p. 114. ISBN : 1581138385

    2004

  2. [10]

    Decision Support and Profit Prediction for Online Auction Sell- ers

    C.-H. Chang and J.-H. Lin. “Decision Support and Profit Prediction for Online Auction Sell- ers”. In: Proceedings of the 1st ACM SIGKDD Workshop on Knowledge Discovery from Uncertain Data. U ’09. Paris, France: Association for Computing Machinery, 2009, pp. 1–8. ISBN : 9781605586755

    2009

  3. [11]

    Healing Sample Selection Bias by Source Classifier Selection

    C.-W. Seah, I. W.-H. Tsang, and Y .-S. Ong. “Healing Sample Selection Bias by Source Classifier Selection”. In: IEEE 11th International Conference on Data Mining . 2011, pp. 577–586. 16

    2011

  4. [12]

    Learning under nonstationarity: covari- ate shift and class-balance change

    M. Sugiyama, M. Y amada, and M. C. du Plessis. “Learning under nonstationarity: covari- ate shift and class-balance change”. In: Wiley Interdisciplinary Reviews: Computational Statistics 5.6 (Aug. 2013), pp. 465–477

    2013

  5. [13]

    Continuous Target Shift Adaptation in Su- pervised Learning

    T. D. Nguyen, M. Christoffel, and M. Sugiyama. “Continuous Target Shift Adaptation in Su- pervised Learning”. In: Asian Conference on Machine Learning . Vol. 45. Proceedings of Machine Learning Research. Hong Kong: PMLR, 20–22 Nov 2016, pp. 285–300

    2016

  6. [14]

    Correcting Sample Selection Bias by Unlabeled Data

    J. Huang, A. J. Smola, A. Gretton, K. M. Borgwardt, and B. Scholkopf. “Correcting Sample Selection Bias by Unlabeled Data”. In: Proceedings of the 19th International Conference on Neural Information Processing Systems. NIPS’06. Canada: MIT Press, 2006, pp. 601–608

    2006

  7. [15]

    Fair and Robust Classification Under Sample Selection Bias

    W. Du and X. Wu. “Fair and Robust Classification Under Sample Selection Bias”. In: Pro- ceedings of the 30th ACM International Conference on Information & Knowledge Manage- ment. CIKM ’21. Virtual Event, Queensland, Australia: Association for Computing Machin- ery, 2021, pp. 2...

    2021

  8. [16]

    Robust classification under sample selection bias

    A. Liu and B. Ziebart. “Robust classification under sample selection bias”. In: Advances in Neural Information Processing Systems 1 (Jan. 2014), pp. 37–45

    2014

  9. [17]

    Robust domain-adaptive discriminant analysis

    W. M. Kouw and M. Loog. “Robust domain-adaptive discriminant analysis”. In: Pattern Recognition Letters 148 (Aug. 2021), pp. 107–113

    2021

  10. [18]

    A survey on semi-supervised learning

    J. E. van Engelen and H. H. Hoos. “A survey on semi-supervised learning”. In: Machine Learning 109.2 (Nov. 2019), pp. 373–440

    2019

  11. [19]

    Pseudo-Label : The Simple and Efficient Semi-Supervised Learning Method for Deep Neural Networks

    D.-H. Lee. “Pseudo-Label : The Simple and Efficient Semi-Supervised Learning Method for Deep Neural Networks”. In: ICML 2013 Workshop : Challenges in Representation Learning (WREPL) (July 2013)

    2013

  12. [20]

    Radhakrishnan, J

    A. Radhakrishnan, J. Davis, Z. Rabin, B. Lewis, M. Scherreik, and R. Ilin. Enhancing Self- Training Methods. 2023

    2023

  13. [21]

    Y . I. Tepeli and J. P . Gonçalves.DCAST: Diverse Class-Aware Self-Training Mitigates Se- lection Bias for Fairer Learning. 2024. arXiv: 2409.20126 [cs.LG]

    2024 arXiv

  14. [22]

    Arazo, D

    E. Arazo, D. Ortego, P . Albert, N. E. O’Connor, and K. McGuinness. Pseudo-Labeling and Confirmation Bias in Deep Semi-Supervised Learning. 2019

    2019

  15. [23]

    Learning a similarity metric discriminatively, with application to face verification

    S. Chopra, R. Hadsell, and Y . LeCun. “Learning a similarity metric discriminatively, with application to face verification”. In:IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05). Vol. 1. 2005, 539–546 vol. 1

    2005

  16. [24]

    Scikit-learn: Machine Learning in Python

    F . Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P . Prettenhofer, R. Weiss, V. Dubourg, et al. “Scikit-learn: Machine Learning in Python”. In: Journal of Machine Learning Research 12 (2011), pp. 2825–2830

    2011

  17. [25]

    Dheeru and E

    D. Dheeru and E. Karra Taniskidou. UCI Machine Learning Repository. 2017

    2017

  18. [26]

    Classification of Pistachio Species Using Im- proved K-NN Classifier

    I. A. Ozkan, M. Koklu, and R. Saraço ˘glu. “Classification of Pistachio Species Using Im- proved K-NN Classifier”. In: Progress in Nutrition 23.2 (July 2021), e2021044. ISSN : 1129- 8723

    2021

  19. [27]

    Determining the Extinguishing Status of Fuel Flames With Sound Wave by Machine Learning Methods

    M. Koklu and Y . S. Taspinar. “Determining the Extinguishing Status of Fuel Flames With Sound Wave by Machine Learning Methods”. In: IEEE Access 9 (2021), pp. 86207–86216. 17

    2021

  20. [28]

    The use of machine learning methods in classification of pumpkin seeds (Cucurbita pepo L.)

    M. Koklu, S. Sarigil, and O. Ozbek. “The use of machine learning methods in classification of pumpkin seeds (Cucurbita pepo L.)” In: Genetic Resources and Crop Evolution 68.7 (June 2021), pp. 2713–2726

    2021

  21. [29]

    Inhibition of Poly(ADP-Ribose) Polymerase in Tumors from BRCA Mutation Carriers

    P . C. Fong, D. S. Boss, T. A. Y ap, A. Tutt, P . Wu, M. Mergui-Roelvink, P . Mortimer, H. Swais- land, A. Lau, M. J. O’Connor, et al. “Inhibition of Poly(ADP-Ribose) Polymerase in Tumors from BRCA Mutation Carriers”. In: New England Journal of Medicine 361.2 (July 2009), pp. 123–134

    2009

  22. [30]

    PARP inhibitor olaparib is safe and effective in patients with BRCA1 and BRCA2 mutations

    L. Hutchinson. “PARP inhibitor olaparib is safe and effective in patients with BRCA1 and BRCA2 mutations”. In: Nature Reviews Clinical Oncology 7.10 (Sept. 2010), p. 549

    2010

  23. [31]

    Harnessing synthetic lethal interactions in anticancer drug discovery

    D. A. Chan and A. J. Giaccia. “Harnessing synthetic lethal interactions in anticancer drug discovery”. In: Nature Reviews Drug Discovery 10.5 (Apr. 2011), pp. 351–64

    2011

  24. [32]

    Overcoming selection bias in synthetic lethality prediction

    C. Seale, Y . Tepeli, and J. P . Gonçalves. “Overcoming selection bias in synthetic lethality prediction”. In: Bioinformatics 38.18 (July 2022). Ed. by K. Borgwardt, pp. 4360–4368

    2022

  25. [33]

    Non-specific chemical in- hibition of the Fanconi anemia pathway sensitizes cancer cells to cisplatin

    C. Jacquemont, J. A. Simon, A. D. D’Andrea, and T. Taniguchi. “Non-specific chemical in- hibition of the Fanconi anemia pathway sensitizes cancer cells to cisplatin”. In: Molecular Cancer 11.1 (2012), p. 26

    2012

  26. [34]

    Genome-wide CRISPR-Cas9 Screens Reveal Loss of Re- dundancy between PKMYT1 and WEE1 in Glioblastoma Stem-like Cells

    C. M. Toledo, Y . Ding, P . Hoellerbauer, R. J. Davis, R. Basom, E. J. Girard, E. Lee, P . Corrin, T. Hart, H. Bolouri, et al. “Genome-wide CRISPR-Cas9 Screens Reveal Loss of Re- dundancy between PKMYT1 and WEE1 in Glioblastoma Stem-like Cells”. In: Cell Reports 13.11 (Dec. 20...

    2015

  27. [35]

    Synthetic lethality between CCNE1 ampli- fication and loss of BRCA1

    D. Etemadmoghadam, B. A. Weir, G. Au-Y eung, K. Alsop, G. Mitchell, J. George, S. Davis, A. D. D’Andrea, K. Simpson, W. C. Hahn, et al. “Synthetic lethality between CCNE1 ampli- fication and loss of BRCA1”. In: Proceedings of the National Academy of Sciences 110.48 (Nov. 2013)...

  28. [36]

    Genome-wide RNAi screens in human brain tumor isolates reveal a novel viability requirement for PHF5A

    C. G. Hubert, R. K. Bradley, Y . Ding, C. M. Toledo, J. Herman, K. Skutt-Kakaria, E. J. Girard, J. Davison, J. Berndt, P . Corrin, et al. “Genome-wide RNAi screens in human brain tumor isolates reveal a novel viability requirement for PHF5A”. In: Genes & Development 27.9 (May ...

    2013

  29. [37]

    A synthetic lethal screen identifies FAT1 as an antagonist of caspase-8 in extrinsic apoptosis

    D. Kranz and M. Boutros. “A synthetic lethal screen identifies FAT1 as an antagonist of caspase-8 in extrinsic apoptosis”. In: The EMBO Journal 33 (Jan. 2014), pp. 181–97

    2014

  30. [38]

    ELISL: early–late integrated synthetic lethality prediction in cancer

    Y . I. Tepeli, C. Seale, and J. P . Gonçalves. “ELISL: early–late integrated synthetic lethality prediction in cancer”. In: Bioinformatics 40.1 (Dec. 2023). Ed. by J. Wren.ISSN : 1367-4811

    2023

  31. [39]

    Harnessing synthetic lethality to predict the response to cancer treatment

    J. S. Lee, A. Das, L. Jerby-Arnon, R. Arafeh, N. Auslander, M. Davidson, L. McGarry, D. James, A. Amzallag, S. G. Park, et al. “Harnessing synthetic lethality to predict the response to cancer treatment”. In: Nature Communications 9.1 (June 2018)

    2018

  32. [40]

    DiscoverSL: an R package for multi-omic data driven prediction of synthetic lethality in cancers

    S. Das, X. Deng, K. Camphausen, and U. Shankavaram. “DiscoverSL: an R package for multi-omic data driven prediction of synthetic lethality in cancers”. In: Bioinformatics 35.4 (July 2018). Ed. by R. Schwartz, pp. 701–702

    2018

  33. [41]

    EXP2SL: A Machine Learning Frame- work for Cell-Line-Specific Synthetic Lethality Prediction

    F . Wan, S. Li, T. Tian, Y . Lei, D. Zhao, and J. Zeng. “EXP2SL: A Machine Learning Frame- work for Cell-Line-Specific Synthetic Lethality Prediction”. In: Frontiers in Pharmacology 11 (Feb. 2020). 18 Contents 1 Supplementary Figures 2 2 Supplementary Tables 3 3 Supplementary ...

    2020 arXiv

Pith tools

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