Pith. sign in

REVIEW 4 major objections 5 minor 65 references

TabDeco: A Comprehensive Contrastive Framework for Decoupled Representations in Tabular Data

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 TabDeco claims that multi-level contrastive losses on decoupled local and global features let a tabular transformer beat XGBoost, CatBoost, and LightGBM on most benchmark datasets.

desk verdict A useful taxonomy of contrastive losses for tabular data, but the headline claim of consistent superiority over GBDTs is contradicted by the paper's own tables and the evaluation protocol is not trustworthy. read the letter →

arxiv 2411.11148 v1 pith:EJIDTOK6 submitted 2024-11-17 cs.LG cs.AI

Suiyao Chen , Jing Wu , Yunxiao Wang , Cheng Ji , Tianpei Xie , Daniel Cociorva , Michael Sharps , Cecile Levasseur
show 1 more author
Hakan Brunzell
This is my paper · ORCID
classification cs.LGcs.AI
keywords contrastivelearningtabulardatarepresentationfeaturedecouplingattentionmechanismself-supervisedgradientboostingAUROC
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

TabDeco sets out to show that contrastive learning can succeed on tabular data when positive and negative pairs are constructed at feature, instance, and batch levels rather than only at the instance level. It splits each encoded sample into a local representation and a global representation, then trains with a supervised loss plus combinations of six contrastive losses. The paper reports that on 11 classification benchmarks the TabDeco variants take the best or second-best score on every dataset, and that they surpass leading gradient-boosting algorithms such as XGBoost, CatBoost, and LightGBM on most of them. A sympathetic reader should care because tabular data dominate applied machine learning, and deep representation learners that beat tuned boosting would change the default tool of choice.

What carries the argument

The load-bearing device is global-local feature decoupling. After feature-level (column) and instance-level (row) attention blocks produce an embedding, a local projector $L(\cdot)$ captures instance-specific and feature-specific detail while a global projector $G(\cdot)$ captures dataset-wide and cross-batch patterns. On these two views the paper defines six contrastive losses—$L_{\mathrm{all}}$, $L_{\mathrm{gg}}$, $L_f$, $L_s$, $L_{\mathrm{fs}}$, and $L_{\mathrm{sf}}$—each built from a global term, a local term, and a cross term, and each comparing similarity matrices at a different granularity (feature, instance, batch, or combinations). This machinery is what lets the paper vary only the pairing structure while keeping the architecture fixed, so observed gains are attributed to the decoupled contrastive objective rather than to added capacity.

What would settle it

Re-run Table 2 with the loss combination and architecture for each dataset fixed before test labels are examined, choosing them only on the 15% validation split, and count how often TabDeco beats XGBoost, CatBoost, and LightGBM; if the average AUROC edge disappears, the paper's central 'consistent superiority' claim is not supported.

Watch

Extended reading notes

Core claim

The central claim is that the weakness of prior contrastive tabular methods lies in how they choose sample pairs, not in the network architecture. TabDeco encodes data through attention over both columns and rows, projects the result into local and global feature vectors, and then applies six contrastive losses that contrast features, instances, and batches in different combinations, with the total loss being the supervised loss plus a weighted contrastive term. The paper argues that this multi-level decoupling produces well-separated, interpretable embeddings, and it reports that the best TabDeco variant is the top performer on 7 of 11 datasets and second-best on the remaining 4, including wins against XGBoost, CatBoost, and LightGBM.

Load-bearing premise

The claim of consistent superiority assumes that selecting, for each dataset, the best of 14 loss combinations and 3 architecture variants after seeing test-set numbers does not inflate the reported performance; were that selection fixed in advance or made only on validation data, the margin over gradient boosting would likely shrink.

Editorial extensions

If this is right

  • A deep tabular model trained with the right contrastive loss can match or exceed tuned gradient boosting on the 11 benchmarks, so tree ensembles are not automatically the safe default for tabular classification.
  • Fine-grained contrastive losses such as $L_{\mathrm{fs}}$, $L_{\mathrm{sf}}$, and $L_{\mathrm{gg}}$ improve on the no-contrastive baseline on 6 to 11 of the 11 datasets, while the simple feature-level and instance-level losses $L_f$ and $L_s$ help on only 0 and 3 datasets.
  • The comparison against the SAINT variants isolates the contribution of the decoupling and contrastive modules, since TabDeco without those modules reduces to the SAINT architecture.
  • TabDeco's reported standard deviations are often smaller than those of the boosting baselines, which the paper reads as evidence of more stable predictions.
  • The framework also reaches the best accuracy on the two multi-class datasets (Volkert and MNIST), indicating the decoupled contrastive objective transfers beyond binary classification.

Reading between the lines

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

  • Beyond the paper's protocol, the per-dataset choice of the best of 14 loss combinations and 3 architecture variants after seeing test-set results may inflate the reported edge over boosting; a validation-based selection protocol would test whether the advantage survives.
  • Beyond the paper, because gains are attributed to pairing structure, a natural extension is to replace the fixed loss zoo with a learned or validation-driven loss selector, which could reduce the 14-combination search to a single training run.
  • Beyond the paper, the same local-global projectors could likely be applied in semi-supervised or transfer settings where labels are scarce, even though the paper only demonstrates supervised training.
  • Beyond the paper, a cheap practical rule suggested by Table 3 is to skip $L_f$ and $L_s$ entirely, since they rarely beat the baseline, and to focus on the fine-grained and cross-batch losses.
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 TabDeco, a supervised contrastive learning framework for tabular data that combines row- and column-level attention (in the style of SAINT) with global-local feature decoupling (inspired by SwitchTab) and a menu of six contrastive losses that can be combined into fourteen loss configurations. The authors evaluate TabDeco and two architectural variants on eleven binary and multiclass tabular datasets, reporting AUROC or accuracy against logistic regression, random forests, XGBoost, CatBoost, LightGBM, MLP, VIME, TabNet, TabTransformer, SwitchTab, and three SAINT variants. The abstract and conclusion claim that TabDeco 'consistently surpasses' leading gradient boosting algorithms across benchmark tasks, and Section 4.2 reports that one TabDeco variant ranks best or second-best on all eleven datasets.

Significance. If the central claim were supported, TabDeco would be a practically valuable contribution to tabular representation learning, since the combination of feature decoupling and multi-level contrastive losses is a plausible and interesting design. The paper also has some strengths: it builds on credible prior components (SAINT, SwitchTab), reports means and standard deviations over ten trials, and provides a relatively broad baseline comparison. However, the empirical evidence presented in the manuscript does not support the headline claim of consistent superiority over gradient boosting, and the evaluation protocol is insufficiently specified to allow the results to be reproduced or trusted. The conceptual contribution is therefore currently obscured by an overclaimed and under-specified evaluation.

major comments (4)
  1. [Abstract; Section 5; Table 2] The claim that TabDeco 'consistently surpasses' XGBoost, CatBoost, and LightGBM is directly contradicted by Table 2. The best TabDeco variant is worse than the best GBDT on Bank (93.34 vs 93.39 by LightGBM), Income (91.26 vs 92.57 by LightGBM), and Spambase (97.97 vs 99.01 by LightGBM). This is 8 of 11 datasets, not consistent superiority. The abstract and conclusion must be revised to reflect the actual results, or the experiments must be redone with a protocol that supports the stronger claim.
  2. [Section 4.3; Table 3] The ablation study reports the best performance for each of 14 contrastive-loss combinations across the three TabDeco variants, but the manuscript never states how the final model used in Table 2 was selected. If the per-dataset best loss combination was chosen after inspecting test-set results in Table 3, the reported Table 2 numbers are optimistically biased by selection on the test set, and the usual safeguards (validation-based selection, multiple-testing correction) are not described. The authors must either disclose the exact selection rule or report results for a single fixed loss configuration chosen before seeing test labels.
  3. [Section 4.2; Tables 2 and 5] The paper describes TabDeco as having 'significant predicting power,' but Table 5 shows that most differences from the strongest GBDT baselines are within about two standard deviations (e.g., Bank TabDeco 93.34 ± 0.15 vs LightGBM 93.39 ± 0.21; Income 91.26 ± 0.07 vs LightGBM 92.57 ± 0.13). No statistical significance test (paired test, confidence interval, or multiple-comparison correction) is reported. Without such an analysis, the claim of consistent superiority is not supported by the evidence in the tables.
  4. [Section 4.1.3; Algorithm 1] The training details specify the optimizer, learning rate, batch size, and temperature, but the contrastive loss weight α and the specific loss combination used for the main results in Table 2 are not given. Algorithm 1 leaves both as free choices (steps 7 and 8). This omission makes the main results non-reproducible and prevents a reader from determining whether the reported performance comes from a fixed model or from per-dataset tuning. The manuscript must specify the exact configuration used to produce each Table 2 entry.
minor comments (5)
  1. [Throughout] There are numerous typos and misspellings, including 'ehance' in Section 1, 'simutaneously' in Section 3.1, 'demoted' before Equation (2), 'constrastive' in Algorithm 1, 'Table‘3' in Section 4.3, 'TabTransormer' in Table 2, and inconsistent spacing in 'V olkert'.
  2. [Section 4.1.5; Reference [49]] The baseline list refers to TabTransformer but cites reference [49], which is the TranTab paper; either the citation or the baseline name is mismatched.
  3. [Equations (2)-(4)] The use of a negative sign inside the exponential for Llocal and Lcross is unusual for InfoNCE-style losses; please clarify whether this is intentional (to repel similar features) or a typo, and relate the formulation to standard contrastive losses.
  4. [Section 1; Section 4.2] The statement that TabDeco is 'the first attempt' to explicitly facilitate structured embeddings through contrasting for tabular data is an overclaim given existing contrastive tabular methods (e.g., SCARF, Recontab); the claim should be softened or restricted to the specific multi-level decoupling setup.
  5. [Section 4.1.5; Table 2] The paper states that baseline results are 'quoted from original papers when possible and reproduced otherwise,' but it does not say which baselines were quoted and which were reproduced, nor whether the same data splits and hyperparameters were used; this should be stated to rule out discrepancies in the comparison.

Circularity Check

1 steps flagged · score 6.0 of 10

TabDeco's headline benchmark wins are the per-dataset best of 14 loss combinations x 3 architectural variants selected on the test metric, so the claimed 'consistent superiority' reduces to post-hoc model selection rather than a fixed-model prediction.

  1. fitted input called prediction [Section 4.3 (Ablation Studies), Table 3; Section 4.2 (Main Results), Table 2; Abstract]
    "We test adding the loss combination to the total loss for each TabDeco variant and report the best performance for each combination from all variants. ... In Table 2, TabDeco demonstrates significant predicting power ranking the best or second best across all datasets. Specifically, in 7 out of 11 datasets, one of the TabDeco variants outperforms all baseline models."

    For each dataset, the best of the three TabDeco rows in Table 2 equals the column-maximum over the 14 loss combinations in Table 3 (e.g., QSARBio 94.87 = Ls; MNIST 97.85 = Lsf+gg). Because Table 3 reports 'the best performance for each combination from all variants' on the test metrics, the loss combination and variant are selected after seeing test results. Reporting that post-hoc maximum as 'TabDeco' and calling it a consistent prediction is equivalent to fitting a discrete hyperparameter to the benchmark. The paper never states which configuration produced each Table 2 entry, whether selection used validation data, or any multiple-testing correction; the superiority claim is therefore forced by the selection protocol, not by a fixed model.

full rationale

The loss definitions, decoupling mechanism, and attention blocks are not circular: they are stated with explicit equations and are not defined in terms of the benchmark outcome, and the paper makes no load-bearing appeal to a self-citation or uniqueness theorem. The one substantive circular step is in evaluation: the headline 'consistent superiority' is built from per-dataset maxima over a 14-combination x 3-variant grid whose values are reported on the test metric in Table 3, with no stated a priori selection rule. That makes the reported performance a fitted quantity, not a prediction of a fixed model. Separately, Table 2 itself shows the best TabDeco variant losing to the best GBDT on Bank (93.34 vs 93.39), Income (91.26 vs 92.57), and Spambase (97.97 vs 99.01), so the abstract's 'consistently surpasses' is also factually overstated; that is a correctness concern rather than a circularity concern.

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

The central claim rests on an evaluation protocol rather than a mathematical derivation. The most important 'free parameter' is the per-dataset choice of loss combination, which is selected from Table 3 after seeing test performance. The paper also leaves alpha (contrastive loss weight) unspecified, hand-sets tau=0.5, and adjusts architecture sizes per dataset. The axioms list the implicit assumptions about baseline comparability, evaluation protocol, and the validity of the under-specified loss equations.

free parameters (4)
  • Per-dataset contrastive loss combination = Varies per dataset (e.g., Lfs+gg, Lall+gg, Lsf+fs+gg), selected as best over 14 combinations in Table 3
    The main results in Table 2 are produced by choosing, for each dataset, the loss combination and variant that performed best in the ablations of Table 3. This is effectively a high-dimensional hyperparameter fit to the benchmark.
  • Contrastive loss weight alpha = Not reported
    Algorithm 1 line 8 defines Ltotal = Lsup + alpha*Lcontrastive, but the paper never states the value of alpha or how it was tuned.
  • Temperature tau = 0.5
    Section 4.1.3 sets tau = 0.5; it is a hand-chosen hyperparameter of the contrastive losses.
  • Architecture sizes per dataset = batch 128 (smaller for MNIST/Volkert), d=32 (d=4/8 for MNIST/Volkert), L=6 (2 for MNIST/Volkert), h=8 (4 for…
    These architectural choices are adjusted per dataset and may affect the reported comparisons.
assumptions (5)
  • domain assumption Baseline numbers from original papers are directly comparable to numbers produced by the TALENT toolbox in this paper
    Section 4.1.5 states baseline results are quoted from original papers when possible and reproduced otherwise, but no protocol is given for merging numbers across different implementations and hardware.
  • domain assumption The evaluation protocol (65/15/20 split, AUROC/accuracy choice, 10 seeds) is sufficient to support the claim of consistent superiority
    Section 4.1.3 and 4.1.4 define the setup, but no significance tests are reported, so differences near one standard deviation are treated as meaningful.
  • ad hoc to paper The contrastive losses defined in Equations (2)-(4) are correctly normalized information-theoretic contrasts
    The paper's notation is ambiguous: Lglobal uses sim(gi,gj) over an unspecified index j, and Table 1 gives shapes (b,m,d) that do not match the per-instance vectors gi and li defined in Eq. (1). The normalization sums over 'the entire set of features' without specifying whether this is over instances, features, or both.
  • domain assumption The TALENT toolbox provides faithful implementations of all baselines
    Section 4.1.3 relies on TALENT for training and comparisons; if a baseline is misconfigured, the relative ranking changes.
  • domain assumption Forest (Covertype) can be evaluated with AUROC despite being a 7-class dataset
    The paper states only Volkert and MNIST are multi-class and use accuracy, yet Table 2 reports AUROC for Forest without describing any adaptation to multiclass AUROC.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TabDeco: A Comprehensive Contrastive Framework for Decoupled Representations in Tabular Data." pith.science (2026). https://pith.science/paper/EJIDTOK6

@misc{pith2026241111148,
  author       = {Pith},
  title        = {Pith review of: TabDeco: A Comprehensive Contrastive Framework for Decoupled Representations in Tabular Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EJIDTOK6}},
  note         = {Machine review of arXiv:2411.11148}
}
read the original abstract

Representation learning is a fundamental aspect of modern artificial intelligence, driving substantial improvements across diverse applications. While selfsupervised contrastive learning has led to significant advancements in fields like computer vision and natural language processing, its adaptation to tabular data presents unique challenges. Traditional approaches often prioritize optimizing model architecture and loss functions but may overlook the crucial task of constructing meaningful positive and negative sample pairs from various perspectives like feature interactions, instance-level patterns and batch-specific contexts. To address these challenges, we introduce TabDeco, a novel method that leverages attention-based encoding strategies across both rows and columns and employs contrastive learning framework to effectively disentangle feature representations at multiple levels, including features, instances and data batches. With the innovative feature decoupling hierarchies, TabDeco consistently surpasses existing deep learning methods and leading gradient boosting algorithms, including XG-Boost, CatBoost, and LightGBM, across various benchmark tasks, underscoring its effectiveness in advancing tabular data representation learning.

Figures

Figures reproduced from arXiv: 2411.11148 by the authors.

Figure 1
Figure 1. An overview of the TabDeco Framework. enhance the model’s ability to capture complex interactions among features and instances, making the most out of the unique characteristics of tabular data. Given a tabular dataset represented by D = {(xi , yi)} N i=1, where each xi is a m-dimensional feature vector with yi as its associated label. N is the total number of samples. As shown in [PITH_FULL_IMAGE:figures/full_fig_… view at source ↗
Figure 2
Figure 2. A demonstration of different levels of contrasts [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

65 extracted references · 43 canonical work pages

  1. [1]

    Tabnet: Attentive interpretable tabular learning

    Sercan Ö Arik and Tomas Pfister. Tabnet: Attentive interpretable tabular learning. In Proceed- ings of the AAAI conference on artificial intelligence, volume 35, pages 6679–6687, 2021

    2021

  2. [2]

    Gradient boosting neural networks: Grownet

    Sarkhan Badirli, Xuanqing Liu, Zhengming Xing, Avradeep Bhowmik, Khoa Doan, and Sathiya S Keerthi. Gradient boosting neural networks: Grownet. arXiv preprint arXiv:2002.07971, 2020

    arXiv 2002

  3. [3]

    Scarf: Self-supervised contrastive learning using random feature corruption

    Dara Bahri, Heinrich Jiang, Yi Tay, and Donald Metzler. Scarf: Self-supervised contrastive learning using random feature corruption. arXiv preprint arXiv:2106.15147, 2021

    arXiv 2021

  4. [4]

    Deep neural networks and tabular data: A survey

    Vadim Borisov, Tobias Leemann, Kathrin Seßler, Johannes Haug, Martin Pawelczyk, and Gjergji Kasneci. Deep neural networks and tabular data: A survey. IEEE Transactions on Neural Networks and Learning Systems, 2022

    work page 2022

  5. [5]

    Random forests

    Leo Breiman. Random forests. Machine learning, 45:5–32, 2001

    2001

  6. [6]

    Classification and regression trees

    Leo Breiman. Classification and regression trees. Routledge, 2017

    2017

  7. [7]

    Some recent advances in design of bayesian binomial reliability demonstration tests

    Suiyao Chen. Some recent advances in design of bayesian binomial reliability demonstration tests. USF Tampa Graduate Theses and Dissertations, 2020

    work page 2020

  8. [8]

    Personalized fall risk assessment for long-term care services improvement

    Suiyao Chen, William D Kearns, James L Fozard, and Mingyang Li. Personalized fall risk assessment for long-term care services improvement. In 2017 Annual Reliability and Maintain- ability Symposium (RAMS), pages 1–7. IEEE, 2017

    work page 2017

Show all 65 references
  1. [9]

    Claims data-driven modeling of hospital time-to-readmission risk with latent heterogeneity

    Suiyao Chen, Nan Kong, Xuxue Sun, Hongdao Meng, and Mingyang Li. Claims data-driven modeling of hospital time-to-readmission risk with latent heterogeneity. Health care manage- ment science, 22:156–179, 2019

    2019

  2. [10]

    Deep representation learning for multi-functional degradation modeling of community-dwelling aging population

    Suiyao Chen, Xinyi Liu, Yulei Li, Jing Wu, and Handong Yao. Deep representation learning for multi-functional degradation modeling of community-dwelling aging population. arXiv preprint arXiv:2404.05613, 2024

    2024 arXiv

  3. [11]

    Multi-state reliability demonstration tests

    Suiyao Chen, Lu Lu, and Mingyang Li. Multi-state reliability demonstration tests. Quality Engineering, 29(3):431–445, 2017

    2017

  4. [12]

    A data heterogeneity modeling and quantification approach for field pre-assessment of chloride-induced corrosion in aging infrastructures

    Suiyao Chen, Lu Lu, Yisha Xiang, Qing Lu, and Mingyang Li. A data heterogeneity modeling and quantification approach for field pre-assessment of chloride-induced corrosion in aging infrastructures. Reliability Engineering & System Safety, 171:123–135, 2018

    2018

  5. [13]

    Recontab: Regularized con- trastive representation learning for tabular data

    Suiyao Chen, Jing Wu, Naira Hovakimyan, and Handong Yao. Recontab: Regularized con- trastive representation learning for tabular data. arXiv preprint arXiv:2310.18541, 2023

    2023 arXiv

  6. [14]

    Xgboost: A scalable tree boosting system

    Tianqi Chen and Carlos Guestrin. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pages 785–794, 2016

    2016

  7. [15]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. In International conference on machine learning, pages 1597–1607. PMLR, 2020

    2020

  8. [16]

    Decoupling zero-shot semantic segmenta- tion

    Jian Ding, Nan Xue, Gui-Song Xia, and Dengxin Dai. Decoupling zero-shot semantic segmenta- tion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11583–11592, 2022

    2022

  9. [17]

    Revisiting deep learning models for tabular data.Advances in Neural Information Processing Systems, 34:18932– 18943, 2021

    Yury Gorishniy, Ivan Rubachev, Valentin Khrulkov, and Artem Babenko. Revisiting deep learning models for tabular data.Advances in Neural Information Processing Systems, 34:18932– 18943, 2021. 9

    2021

  10. [18]

    Analysis of the automl challenge series 2015-2018

    Isabelle Guyon, Lisheng Sun-Hosoya, Marc Boullé, Hugo Jair Escalante, Sergio Escalera, Zhengying Liu, Damir Jajetic, Bisakha Ray, Mehreen Saeed, Michéle Sebag, Alexander Stat- nikov, WeiWei Tu, and Evelyne Viegas. Analysis of the automl challenge series 2015-2018. In AutoML, S...

    2015

  11. [19]

    Generalized linear models

    Trevor J Hastie and Daryl Pregibon. Generalized linear models. In Statistical models in S, pages 195–247. Routledge, 2017

    2017

  12. [20]

    Momentum contrast for unsupervised visual representation learning

    Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. Momentum contrast for unsupervised visual representation learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9729–9738, 2020

    2020

  13. [21]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016

    2016

  14. [22]

    Tabtransformer: Tabular data modeling using contextual embeddings

    Xin Huang, Ashish Khetan, Milan Cvitkovic, and Zohar Karnin. Tabtransformer: Tabular data modeling using contextual embeddings. arXiv preprint arXiv:2012.06678, 2020

    2012 arXiv

  15. [23]

    Lightgbm: A highly efficient gradient boosting decision tree

    Guolin Ke, Qi Meng, Thomas Finley, Taifeng Wang, Wei Chen, Weidong Ma, Qiwei Ye, and Tie-Yan Liu. Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems, 30, 2017

    2017

  16. [24]

    Deepgbm: A deep learning framework distilled by gbdt for online prediction tasks

    Guolin Ke, Zhenhui Xu, Jia Zhang, Jiang Bian, and Tie-Yan Liu. Deepgbm: A deep learning framework distilled by gbdt for online prediction tasks. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 384–394, 2019

    2019

  17. [25]

    Tabnn: A universal neural network solution for tabular data

    Guolin Ke, Jia Zhang, Zhenhui Xu, Jiang Bian, and Tie-Yan Liu. Tabnn: A universal neural network solution for tabular data. 2018

    2018

  18. [26]

    Supervised contrastive learning

    Prannay Khosla, Piotr Teterwak, Chen Wang, Abhinav Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. Supervised contrastive learning. Advances in Neural Information Processing Systems, 33:18661–18673, 2020

    2020

  19. [27]

    Self-normalizing neural networks

    Günter Klambauer, Thomas Unterthiner, Andreas Mayr, and Sepp Hochreiter. Self-normalizing neural networks. Advances in neural information processing systems, 30, 2017

    2017

  20. [28]

    A combinatorial algorithm for approxi- mating the optimal transport in the parallel and mpc settings

    Nathaniel Lahn, Sharath Raghvendra, and Kaiyi Zhang. A combinatorial algorithm for approxi- mating the optimal transport in the parallel and mpc settings. Advances in Neural Information Processing Systems, 36:21675–21686, 2023

    2023

  21. [29]

    Residual-based lan- guage models are free boosters for biomedical imaging tasks

    Zhixin Lai, Jing Wu, Suiyao Chen, Yucheng Zhou, and Naira Hovakimyan. Residual-based lan- guage models are free boosters for biomedical imaging tasks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5086–5096, 2024

    2024

  22. [30]

    Adaptive ensembles of fine-tuned transformers for llm-generated text detection

    Zhixin Lai, Xuesheng Zhang, and Suiyao Chen. Adaptive ensembles of fine-tuned transformers for llm-generated text detection. arXiv preprint arXiv:2403.13335, 2024

    2024 arXiv

  23. [31]

    Machine learning in agriculture: A review

    Konstantinos G Liakos, Patrizia Busato, Dimitrios Moshou, Simon Pearson, and Dionysis Bochtis. Machine learning in agriculture: A review. Sensors, 18(8):2674, 2018

    2018

  24. [32]

    Talent: A tabular analytics and learning toolbox

    Si-Yang Liu, Hao-Run Cai, Qi-Le Zhou, and Han-Jia Ye. Talent: A tabular analytics and learning toolbox. arXiv preprint arXiv:2407.04057, 2024

    2024 arXiv

  25. [33]

    An efficient framework for learning sentence repre- sentations

    Lajanugen Logeswaran and Honglak Lee. An efficient framework for learning sentence repre- sentations. In International Conference on Learning Representations, 2018

    2018

  26. [34]

    Unsupervised domain-specific deblurring via disentangled representations

    Boyu Lu, Jun-Cheng Chen, and Rama Chellappa. Unsupervised domain-specific deblurring via disentangled representations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10225–10234, 2019

    2019

  27. [35]

    A data-driven approach to predict the success of bank telemarketing

    Sérgio Moro, Paulo Cortez, and Paulo Rita. A data-driven approach to predict the success of bank telemarketing. Decision Support Systems, 62:22–31, 2014

    2014

  28. [36]

    Telco customer churn

    James Ouk, David Dada, and Kyung Tae Kang. Telco customer churn. 2018

    2018

  29. [37]

    Computing all optimal partial transports

    Abhijeet Phatak, Sharath Raghvendra, Chittaranjan Tripathy, and Kaiyi Zhang. Computing all optimal partial transports. In International Conference on Learning Representations, 2023

    2023

  30. [38]

    Neural oblivious decision ensembles for deep learning on tabular data

    Sergei Popov, Stanislav Morozov, and Artem Babenko. Neural oblivious decision ensembles for deep learning on tabular data. arXiv preprint arXiv:1909.06312, 2019. 10

    1909 arXiv

  31. [39]

    Catboost: unbiased boosting with categorical features

    Liudmila Prokhorenkova, Gleb Gusev, Aleksandr V orobev, Anna Veronika Dorogush, and Andrey Gulin. Catboost: unbiased boosting with categorical features. Advances in neural information processing systems, 31, 2018

    2018

  32. [40]

    Secure and robust machine learning for healthcare: A survey

    Adnan Qayyum, Junaid Qadir, Muhammad Bilal, and Ala Al-Fuqaha. Secure and robust machine learning for healthcare: A survey. IEEE Reviews in Biomedical Engineering, 14:156– 180, 2020

    2020

  33. [41]

    A new robust partial p-wasserstein- based metric for comparing distributions

    Sharath Raghvendra, Pouyan Shirzadian, and Kaiyi Zhang. A new robust partial p-wasserstein- based metric for comparing distributions. In International Conference on Machine Learning, 2024

    2024

  34. [42]

    Real-time prediction of online shoppers’ purchasing intention using multilayer perceptron and lstm recurrent neural networks

    C Okan Sakar, S Olcay Polat, Mete Katircioglu, and Yomi Kastro. Real-time prediction of online shoppers’ purchasing intention using multilayer perceptron and lstm recurrent neural networks. Neural Computing and Applications, 31:6893–6908, 2019

    2019

  35. [43]

    Saint: Improved neural networks for tabular data via row attention and contrastive pre-training

    Gowthami Somepalli, Micah Goldblum, Avi Schwarzschild, C Bayan Bruss, and Tom Goldstein. Saint: Improved neural networks for tabular data via row attention and contrastive pre-training. arXiv preprint arXiv:2106.01342, 2021

    2021 arXiv

  36. [44]

    Autoint: Automatic feature interaction learning via self-attentive neural networks

    Weiping Song, Chence Shi, Zhiping Xiao, Zhijian Duan, Yewen Xu, Ming Zhang, and Jian Tang. Autoint: Automatic feature interaction learning via self-attentive neural networks. In Proceedings of the 28th ACM international conference on information and knowledge management, pages...

    2019

  37. [45]

    Optimizing crop management with reinforcement learning and imitation learning

    Ran Tao, Pan Zhao, Jing Wu, Nicolas F Martin, Matthew T Harrison, Carla Ferreira, Zahra Kalantari, and Naira Hovakimyan. Optimizing crop management with reinforcement learning and imitation learning. arXiv preprint arXiv:2209.09991, 2022

    2022 arXiv

  38. [46]

    Optimal test design for reliability demonstration under multi-stage acceptance uncertainties

    Bingjie Wang, Lu Lu, Suiyao Chen, and Mingyang Li. Optimal test design for reliability demonstration under multi-stage acceptance uncertainties. Quality Engineering, 0(0):1–14, 2023

    2023

  39. [47]

    Dcn v2: Improved deep & cross network and practical lessons for web-scale learning to rank systems

    Ruoxi Wang, Rakesh Shivanna, Derek Cheng, Sagar Jain, Dong Lin, Lichan Hong, and Ed Chi. Dcn v2: Improved deep & cross network and practical lessons for web-scale learning to rank systems. In Proceedings of the web conference 2021, pages 1785–1797, 2021

    2021

  40. [48]

    Balanced training for sparse gans

    Yite Wang, Jing Wu, Naira Hovakimyan, and Ruoyu Sun. Balanced training for sparse gans. In Thirty-seventh Conference on Neural Information Processing Systems, 2023

    2023

  41. [49]

    Transtab: Learning transferable tabular transformers across tables

    Zifeng Wang and Jimeng Sun. Transtab: Learning transferable tabular transformers across tables. Advances in Neural Information Processing Systems, 35:2902–2915, 2022

    2022

  42. [50]

    Logistic regression

    Raymond E Wright. Logistic regression. 1995

    1995

  43. [51]

    Switchtab: Switched autoencoders are effective tabular learners

    Jing Wu, Suiyao Chen, Qi Zhao, Renat Sergazinov, Chen Li, Shengjie Liu, Chongchao Zhao, Tianpei Xie, Hanqing Guo, Cheng Ji, et al. Switchtab: Switched autoencoders are effective tabular learners. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, page...

    2024

  44. [52]

    Hallucination improves the performance of unsupervised visual representation learning

    Jing Wu, Jennifer Hobbs, and Naira Hovakimyan. Hallucination improves the performance of unsupervised visual representation learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 16132–16143, 2023

    2023

  45. [53]

    Genco: An auxiliary generator from contrastive learning for enhanced few-shot learning in remote sensing

    Jing Wu, Naira Hovakimyan, and Jennifer Hobbs. Genco: An auxiliary generator from contrastive learning for enhanced few-shot learning in remote sensing. arXiv preprint arXiv:2307.14612, 2023

    2023 arXiv

  46. [54]

    The new agronomists: Language models are experts in crop management

    Jing Wu, Zhixin Lai, Suiyao Chen, Ran Tao, Pan Zhao, and Naira Hovakimyan. The new agronomists: Language models are experts in crop management. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5346–5356, 2024

    2024

  47. [55]

    Crops: A deployable crop management system over all possible state availabilities

    Jing Wu, Zhixin Lai, Shengjie Liu, Suiyao Chen, Ran Tao, Pan Zhao, Chuyuan Tao, Yikun Cheng, and Naira Hovakimyan. Crops: A deployable crop management system over all possible state availabilities. arXiv preprint arXiv:2411.06034, 2024

    2024 arXiv

  48. [56]

    Extended agriculture-vision: An extension of a large aerial image dataset for agricultural pattern analysis

    Jing Wu, David Pichler, Daniel Marley, David Wilson, Naira Hovakimyan, and Jennifer Hobbs. Extended agriculture-vision: An extension of a large aerial image dataset for agricultural pattern analysis. arXiv preprint arXiv:2303.02460, 2023. 11

    2023 arXiv

  49. [57]

    Optimizing nitrogen management with deep reinforcement learning and crop simulations

    Jing Wu, Ran Tao, Pan Zhao, Nicolas F Martin, and Naira Hovakimyan. Optimizing nitrogen management with deep reinforcement learning and crop simulations. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 1712–1720, 2022

    2022

  50. [58]

    Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms

    Han Xiao, Kashif Rasul, and Roland V ollgraf. Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747, 2017

    2017 arXiv

  51. [59]

    Multiplexed orbital angular momentum beams demultiplexing using hybrid optical-electronic convolutional neural network

    Jiachi Ye, Haoyan Kang, Qian Cai, Zibo Hu, Maria Solyanik-Gorgone, Hao Wang, Elham Heidari, Chandraman Patil, Mohammad-Ali Miri, Navid Asadizanjani, et al. Multiplexed orbital angular momentum beams demultiplexing using hybrid optical-electronic convolutional neural network. N...

    2024

  52. [60]

    Multiplexed oam beams classification via fourier optical convolutional neural network

    Jiachi Ye, Haoyan Kang, Hao Wang, Salem Altaleb, Elham Heidari, Navid Asadizanjani, V olker J Sorger, and Hamed Dalir. Multiplexed oam beams classification via fourier optical convolutional neural network. In 2023 IEEE Photonics Conference (IPC), pages 1–2. IEEE, 2023

    2023

  53. [61]

    Oam beams multiplexing and classification under atmospheric turbulence via fourier convolutional neural network

    Jiachi Ye, Haoyan Kang, Hao Wang, Salem Altaleb, Elham Heidari, Navid Asadizanjani, V olker J Sorger, and Hamed Dalir. Oam beams multiplexing and classification under atmospheric turbulence via fourier convolutional neural network. In Frontiers in Optics, pages JTu4A–73. Optic...

    2023

  54. [62]

    Demultiplexing oam beams via fourier optical convolutional neural network

    Jiachi Ye, Haoyan Kang, Hao Wang, Chen Shen, Belal Jahannia, Elham Heidari, Navid Asadizanjani, Mohammad-Ali Miri, V olker J Sorger, and Hamed Dalir. Demultiplexing oam beams via fourier optical convolutional neural network. In Laser Beam Shaping XXIII, volume 12667, pages 16–...

    2023

  55. [63]

    Free-space optical multiplexed orbital angular momentum beam identification system using fourier optical convolutional layer based on 4f system

    Jiachi Ye, Maria Solyanik, Zibo Hu, Hamed Dalir, Behrouz Movahhed Nouri, and V olker J Sorger. Free-space optical multiplexed orbital angular momentum beam identification system using fourier optical convolutional layer based on 4f system. In Complex Light and Optical Forces X...

    2023

  56. [64]

    Disentangled non-local neural networks

    Minghao Yin, Zhuliang Yao, Yue Cao, Xiu Li, Zheng Zhang, Stephen Lin, and Han Hu. Disentangled non-local neural networks. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XV 16 , pages 191–207. Springer, 2020

    2020

  57. [65]

    Vime: Extending the success of self-and semi-supervised learning to tabular domain

    Jinsung Yoon, Yao Zhang, James Jordon, and Mihaela van der Schaar. Vime: Extending the success of self-and semi-supervised learning to tabular domain. Advances in Neural Information Processing Systems, 33:11033–11043, 2020. A Appendix / supplemental material A.1 Datasets Detai...

    2020

Pith tools

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