Pith. sign in

REVIEW 4 major objections 5 minor 35 references

VisTabNet: Adapting Vision Transformers for Tabular Data

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read VisTabNet claims that a Vision Transformer encoder pre-trained on images, fed with learned projections of tabular rows, beats tree ensembles and deep tabular models on small tabular datasets.

desk verdict The cross-modal transfer idea is fresh and the benchmark is solid, but the paper never runs the pre-trained-vs-from-scratch control its own definition requires, so the headline transferability claim is unsupported. read the letter →

arxiv 2501.00057 v2 pith:LSATVJMW submitted 2024-12-28 cs.LG cs.AIcs.CV

Witold Wydmański , Ulvi Movsum-zada , Jacek Tabor , Marek Śmieja This is my paper · ORCID
classification cs.LGcs.AIcs.CV
keywords cross-modaltransferlearningtabulardataVisionTransformersmalldatasetsadapternetworkfew-shotMatthewscorrelationcoefficienttransferability
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

VisTabNet tries to establish that a Vision Transformer encoder pre-trained on images can be transplanted across modalities to tabular data. The method replaces ViT's patch embedding layer with a small adapter network that projects each tabular row into several views in ViT's embedding space, then feeds those views through the frozen image-trained encoder to a classification head. On 20 small tabular datasets with fewer than 1,000 samples each, the paper reports the highest mean Matthews correlation coefficient, 67.43, edging out random forests by 1.62 points and the best deep tabular baseline, NODE, by 2.5 points. If the claim holds, pre-trained image models become reusable for tabular problems without designing a tabular architecture or paying the cost of training a large network from scratch.

What carries the argument

The carrying mechanism is the adaptation network $\pi$, a set of $n$ small feed-forward projections that maps a tabular row into $n$ vectors in the same $D$-dimensional space ViT uses for image patch embeddings. These projected views play the role of patch embeddings, with a CLS token prepended, so the frozen image-pretrained Transformer Encoder can process tabular data unchanged. The mechanism is supposed to make tabular inputs and image patch inputs similar enough in representation space that the pre-trained attention layers transfer useful structure; conceptually the paper frames this as forcing similarity, while in practice the adapter is trained by classification error alone. The choice of which ViT layers to keep, the backbone size, and whether to fine-tune the encoder are analyzed as secondary controls on the same mechanism.

What would settle it

Train VisTabNet with the identical adapter and head but a randomly initialized ViT encoder, holding layer choice, learning rate, and tuning budget fixed; if the random-encoder version matches the pre-trained version's MCC across the 20 benchmark datasets, the pre-trained image weights are not what carries the result.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that transferability of a pre-trained network is not limited to early feature extractors and is not confined to the original data modality. The paper defines transferability as follows: a pre-trained encoder $g_\theta$ is transferable to a downstream task if a new head $h'_\phi$ composed with it performs at least as well as the same architecture trained from scratch. VisTabNet realizes this by replacing ViT's patch embedding layer $f$ with an adaptation network $\pi = (\pi_1,\ldots,\pi_n)$, where each $\pi_i$ is a feed-forward map from the tabular input $x \in \mathbb{R}^M$ to a view $v_i \in \mathbb{R}^D$ analogous to a patch embedding. The sequence $[\mathrm{CLS}, v_1, \ldots, v_n]$ is then processed by the frozen pre-trained ViT encoder, and an MLP head produces the prediction. The experiments report the best mean MCC and best mean rank against random forests, gradient boosting, XGBoost, LightGBM, ResNet, Feature Transformer, and NODE across 20 datasets, and improved few-shot performance from 2 to 10 examples per class. The paper also reports that starting the encoder at an intermediate layer (around layer 5) improves average MCC over using all 12 layers, and that fully training all parameters at once degrades performance.

Load-bearing premise

The load-bearing premise is that a Transformer encoder trained on images keeps supplying useful representations when its inputs are learned projections of tabular rows, rather than the adapter and classification head doing all of the work.

Editorial extensions

If this is right

  • Cross-modal transfer is viable: an encoder trained on images can serve tabular tasks, so pre-trained model libraries are not locked to their original modality.
  • Middle layers of a pre-trained transformer are transferable, not just early feature extractors; the paper reports that starting from layer 5 of ViT outperforms using the full encoder.
  • Small tabular datasets can benefit from large pre-trained models with few trainable parameters, reducing both overfitting and training cost relative to deep models trained from scratch.
  • The same adapter idea extends in part to NLP encoders: BERT-based versions are promising, while the ViT-based version achieves the best mean rank.
  • The stable default is to freeze the encoder and train only the adapter and head; training all parameters at once clearly hurts performance.

Reading between the lines

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

  • Beyond the paper, a decisive comparison would pit VisTabNet with pre-trained ViT weights against VisTabNet with a randomly initialized ViT encoder of the same architecture; the paper's ablations remove the encoder entirely but do not isolate pre-training as the cause.
  • Beyond the paper, if the gain comes from generic attention over learned views rather than image-specific features, then other sequence encoders, such as time-series or graph transformers, should transfer with the same adapter trick, which is a testable prediction.
  • Beyond the paper, the benchmark covers only small datasets; extending VisTabNet to larger tabular benchmarks with high-cardinality categorical features would show where tree ensembles regain the edge.
  • Beyond the paper, the paper's similarity-forcing language suggests an explicit distribution-matching loss between projected views and real patch embeddings; adding such a loss and measuring MCC changes would directly test the mechanism the paper invokes.
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 VisTabNet, a cross-modal transfer learning method that replaces the patch embedding layer of a pre-trained Vision Transformer (ViT) with an adaptation network of feed-forward projections. Each projection maps a tabular row into a vector in the ViT patch-embedding space, and the resulting sequence of views is processed by the frozen ViT encoder before a classification head. The method is benchmarked on 23 small tabular datasets (the text says 20) against random forests, gradient boosting, XGBoost, LightGBM, ResNet, Feature Transformer, and NODE, using double cross-validation, identical splits, and three repeats. The paper also reports few-shot experiments and ablations over backbone choice, fine-tuning strategy, depth of adapter/head, and the selection of ViT encoder layers. The central claim is that pre-trained image models can be transferred to tabular data and outperform both tree ensembles and tabular deep learning models on small datasets.

Significance. The idea of adapting a pre-trained vision transformer to tabular data is original and practically appealing, since it promises to reuse large-scale image representations without designing task-specific architectures or training from scratch. The experimental protocol has clear strengths: double cross-validation, identical train/validation/test splits across methods, hyperparameter optimization with PyHopper, three repeated runs, the use of MCC for class-imbalanced data, and a public code repository. If the central transferability claim were convincingly established, the result would extend transfer learning beyond same-domain feature reuse and would be of broad interest. However, the paper's own transferability definition is not tested by the provided experiments, because the pre-trained ViT is never compared against the same architecture trained from scratch; and the headline improvements over random forests are not shown to be statistically significant. The work therefore currently establishes a promising architecture and benchmark but not the cross-modal transfer phenomenon as defined in the paper.

major comments (4)
  1. [Section 3.3, Table 2] The transferability definition in Section 3.2 requires that a pre-trained network g_theta with adapter/head performs at least as well as the same architecture trained from scratch. Table 2 instead compares VisTabNet with a 'No ViT encoder' variant, which removes the transformer body entirely, thereby changing the architecture (depth, attention mechanism, parameter count). Any performance difference between these two configurations cannot be attributed to the pre-trained initialization. Therefore the statement in Section 3.3 that 'this property holds in most cases for ViT (Table 2)' is not supported by the presented experiment. A baseline with a randomly initialized ViT encoder of the same architecture (same adapter and head, same training protocol) is needed to test the actual cross-modal transfer claim.
  2. [Section 3.3, Conclusion] The paper repeatedly states that VisTabNet works by 'forcing similarity between transformed tabular inputs and the embeddings of image patches' (Section 3.3 and Conclusion). However, no similarity loss, distribution-alignment term, or any auxiliary objective is defined or trained; the adapter and head are trained exclusively with classification error, as described in Section 3.3 and the hyperparameter details in Appendix A. Consequently, the proposed mechanism is not implemented or tested. Either add an explicit alignment objective (and demonstrate its effect) or rephrase these statements to say that the adapter is designed to produce inputs in patch-embedding space, not that similarity is actively enforced.
  3. [Section 4.1, Table 1] The headline claim of superiority rests on a mean MCC difference of 1.62 points over Random Forests (67.43 vs. 65.81) and a mean rank difference of 0.11 (3.93 vs. 4.04), with standard deviations that overlap across datasets. No significance test (e.g., Wilcoxon signed-rank test or paired permutation test over dataset means) is reported. Given the small magnitude of the difference and the variability visible in Table 1, the claim of overall superiority is not statistically supported. Add paired significance tests across the datasets and report p-values or bootstrap confidence intervals for both MCC differences and ranks.
  4. [Section 4.1, Figure 2] The few-shot experiment claims that VisTabNet 'achieves significantly better scores' and 'consistently outperforms' other methods between 2 and 10 shots. The evidence is based on only 5 datasets, no error bars are shown, and no significance test is reported; moreover, for the 1-shot case RF and GB outperform VisTabNet. Report per-shot variability (e.g., across the three repeats or multiple seeds) and apply a paired test, or temper the claim to state an average improvement over the 2-10 shot range without the word 'significantly'.
minor comments (5)
  1. [Section 4.1, Appendix A] The text states that 'The experiments were conducted on 20 datasets', but Table 4 lists 23 datasets. Please correct the count.
  2. [Abstract] The abstract says datasets with 'less than 1k samples', yet Statlog German has exactly 1000 samples. Use 'at most 1k' or adjust the dataset description.
  3. [References] References [9] and [28] contain bracketed editorial annotations such as '99 citations (Semantic Scholar/arXiv) [2023-02-06]' and '561 citations (Semantic Scholar/arXiv) [2024-01-16]'. These appear to be leftover notes and should be removed.
  4. [Section 4.2] There are several typographical errors, including 'paramterized', 'learnig rate', and 'V it' in the text around Section 4.2; a careful copyediting pass is needed.
  5. [Table 2] Unlike Table 1, Table 2 does not report standard deviations for the MCC values. Adding them would help assess whether the differences between fine-tuning strategies and backbones are meaningful.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central results are empirical benchmarks against external baselines, and the only self-citation is background rather than load-bearing.

full rationale

The paper makes an empirical architecture proposal and validates it with external benchmarks, so there is no derivation chain that reduces to its own inputs. The transferability definition in Section 3.2 is used only to frame the experiments; the results in Table 2 are presented as evidence, and although the 'No ViT encoder' ablation does not instantiate that definition because the transformer body is removed rather than re-initialized from scratch, this is an experimental validity gap rather than a circular reduction because the reported MCC values are not constructed from the claim being made. The only self-citation, HyperTab [32], appears in the related-work inventory and is never used to justify VisTabNet's mechanism or results. No parameter is fitted to a target and then reported as a prediction of that same target, no uniqueness theorem is imported from the authors' prior work, and the 'forcing similarity' language in Sections 3.3 and 5 describes an intended mechanism that is not implemented as an auxiliary loss, which is again an empirical or soundness concern rather than a self-referential derivation. Accordingly, no step exhibits Eq. X = Eq. Y by construction or a fitted-input-renamed-as-prediction pattern.

Assumptions & free parameters 6 free parameters · 4 assumptions · 1 invented entities

The central claim rests on standard empirical ML assumptions (representative small datasets, adequate hyperparameter tuning, MCC as metric) plus the paper-specific assumption that a frozen ImageNet ViT encoder is useful for tabular inputs without explicit alignment. The main free parameters are per-dataset hyperparameters and a post-hoc layer range selection.

free parameters (6)
  • Number of projections (PROJECTIONS) = Chosen per dataset from {8, 16, 32, 64, 128} via hyperparameter optimization.
    Controls the number of learned views fed to the ViT encoder; is a per-dataset hyperparameter.
  • Projection depth (PROJ_DEPTH) = Chosen per dataset from {1, 2, 3, 4} via hyperparameter optimization.
    Depth of each feed-forward view projector; tuned per dataset.
  • Learning rates (LR, PROJ_LR) = Chosen per dataset in the range 1e-5 to 1e-3.
    Separate learning rates for the head and the adapter; tuned per dataset.
  • Epochs = Chosen per dataset from 10 to 100 (multiples of 10).
    Training length; tuned per dataset.
  • Encoder layer range (start, end) = Best start=5, end=12 on the 5 analysis datasets.
    Selected post-hoc on validation performance; the paper reports 0.75 MCC for full encoder versus about 0.78 for layers 5-12.
  • Fine-tuning learning rate = Manually selected by the authors.
    The paper notes manual learning rate selection was needed to stabilize fine-tuning results.
assumptions (4)
  • ad hoc to paper A ViT encoder pre-trained on ImageNet representations is transferable to tabular data without any explicit distribution-alignment training.
    Section 3.3 states that forcing similarity between tabular views and patch embeddings leads to transferability, but no loss enforces this; the assumption is tested only indirectly by the 'No ViT encoder' ablation.
  • domain assumption The tabular datasets used (all under 1,000 samples) are representative of the small-data regime where the method is claimed to be superior.
    Section 4.1 explicitly restricts to small datasets and states VisTabNet performs best there; the paper does not evaluate larger datasets.
  • domain assumption MCC is a sufficient performance metric for comparing classifiers on imbalanced small tabular datasets.
    Section 4.1 justifies MCC as robust to imbalance; no alternative metrics are reported.
  • domain assumption Hyperparameter optimization with 50 PyHopper steps and three repeats provides a fair comparison across methods.
    Appendix A; the budget is modest, and no statistical tests are applied to the final comparisons.
invented entities (1)
  • Adaptation network pi (multiple feed-forward projections)
    purpose: Maps tabular rows to ViT-compatible patch embeddings (views) to enable cross-modal transfer into the frozen ViT encoder.
    The adapter is a new architectural component introduced in this paper; its effectiveness is supported only by the paper's own experiments, with no external validation or theoretical guarantee.

how reviews work

0 comments
Cite this review

Pith. "Pith review of VisTabNet: Adapting Vision Transformers for Tabular Data." pith.science (2026). https://pith.science/paper/LSATVJMW

@misc{pith2026250100057,
  author       = {Pith},
  title        = {Pith review of: VisTabNet: Adapting Vision Transformers for Tabular Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LSATVJMW}},
  note         = {Machine review of arXiv:2501.00057}
}
read the original abstract

Although deep learning models have had great success in natural language processing and computer vision, we do not observe comparable improvements in the case of tabular data, which is still the most common data type used in biological, industrial and financial applications. In particular, it is challenging to transfer large-scale pre-trained models to downstream tasks defined on small tabular datasets. To address this, we propose VisTabNet -- a cross-modal transfer learning method, which allows for adapting Vision Transformer (ViT) with pre-trained weights to process tabular data. By projecting tabular inputs to patch embeddings acceptable by ViT, we can directly apply a pre-trained Transformer Encoder to tabular inputs. This approach eliminates the conceptual cost of designing a suitable architecture for processing tabular data, while reducing the computational cost of training the model from scratch. Experimental results on multiple small tabular datasets (less than 1k samples) demonstrate VisTabNet's superiority, outperforming both traditional ensemble methods and recent deep learning models. The proposed method goes beyond conventional transfer learning practice and shows that pre-trained image models can be transferred to solve tabular problems, extending the boundaries of transfer learning. We share our example implementation as a GitHub repository available at https://github.com/wwydmanski/VisTabNet.

Figures

Figures reproduced from arXiv: 2501.00057 by the authors.

Figure 1
Figure 1. Data flow architecture in VisTabNet. The [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Average performance on 5 datasets (Credit [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Influence of the depth of adaptation and [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Average performance of VisTabNet when selected layers were removed from the ViT encoder using 5 datasets (ZOO, Dermatology, Credit Approval, Cylinder Bands, Libras). The heatmap presented in [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Performance of VisTabNet when selected layers were removed from the ViT encoder. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Learning curves across training epochs of VisTabNet. Red color indicates the phase of training [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 27 canonical work pages

  1. [1]

    S. ¨O. Arik and T. Pfister , Tabnet: Attentive interpretable tabular learning, CoRR, abs/1908.07442 (2019). Submitted to SIAM Conference on Data Mining

    arXiv 2019

  2. [2]

    Breiman , Random forests, Machine Learning, 45 (2001), pp

    L. Breiman , Random forests, Machine Learning, 45 (2001), pp. 5–32

    work page 2001

  3. [3]

    Chen and C

    T. Chen and C. Guestrin , Xgboost: A scalable tree boosting system, Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Dis- covery and Data Mining, (2016)

    work page 2016

  4. [4]

    Chicco, N

    D. Chicco, N. T ¨otsch, and G. Jurman , The Matthews correlation coefficient (MCC) is more reliable than balanced accuracy, bookmaker informedness, and markedness in two-class confusion matrix evaluation, BioData Mining, 14 (2021), p. 13

    work page 2021

  5. [5]

    Chowdhury and et al

    R. Chowdhury and et al. , Predicting the stock price of frontier markets using machine learning and modified black–scholes option pricing model, Physica A: Statistical Mechanics and its Applications, 555 (2020), p. 124444

    work page 2020

  6. [6]

    Cortes and V

    C. Cortes and V. V apnik, Support-vector networks, Machine learning, 20 (1995), pp. 273–297

    work page 1995

  7. [7]

    Dosovitskiy, L

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby , An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale, June 2021. arXiv:2010.11929 [cs]

    arXiv 2021

  8. [8]

    J. H. Friedman , Greedy function approximation: A gradient boosting machine., Annals of Statistics, 29 (2001), pp. 1189–1232

    work page 2001

Show all 35 references
  1. [9]

    Gorishniy, I

    Y. Gorishniy, I. Rubachev, V. Khrulkov, and A. Babenko , Revisiting Deep Learning Models for Tabular Data, Nov. 2021. 99 citations (Semantic Scholar/arXiv) [2023-02-06] arXiv:2106.11959 [cs]

    2021 arXiv

  2. [10]

    Grinsztajn, E

    L. Grinsztajn, E. Oyallon, and G. V aroquaux , Why do tree-based models still outperform deep learning on tabular data?, p. 34

  3. [11]

    Grinsztajn, E

    L. Grinsztajn, E. Oyallon, and G. V aroquaux , Why do tree-based models still outperform deep learning on typical tabular data?, Advances in neural information processing systems, 35 (2022), pp. 507– 520

    2022

  4. [12]

    Kadra, M

    A. Kadra, M. Lindauer, F. Hutter, and J. Grabocka, Regularization is all you need: Simple neural nets can excel on tabular data, CoRR, abs/2106.11189 (2021)

    2021 arXiv

  5. [13]

    Katzir, G

    L. Katzir, G. Elidan, and R. El-Yaniv, Net-{dnf}: Effective deep modeling of tabular data, in Interna- tional Conference on Learning Representations, 2021

    2021

  6. [14]

    G. Ke, Q. Meng, T. Finley, T. W ang, W. Chen, W. Ma, Q. Ye, and T.-Y. Liu , Lightgbm: A highly efficient gradient boosting decision tree, Ad- vances in neural information processing systems, 30 (2017), pp. 3146–3154

    2017

  7. [15]

    Kossen and et al

    J. Kossen and et al. , Self-attention between datapoints: Going beyond individual input-output pairs in deep learning, in NeurIPS, 2021

    2021

  8. [16]

    Krizhevsky, I

    A. Krizhevsky, I. Sutskever, and G. E. Hinton , ImageNet classification with deep convolutional neural networks, Communications of the ACM, 60 (2017), pp. 84–90

    2017

  9. [17]

    Lhoest, A

    Q. Lhoest, A. Villanova del Moral, Y. Jernite, A. Thakur, P. von Platen, S. Patil, J. Chau- mond, M. Drame, J. Plu, L. Tunstall, J. Davi- son, M. ˇSaˇsko, G. Chhablani, B. Malik, S. Bran- deis, T. Le Scao, V. Sanh, C. Xu, N. Pa- try, A. McMillan-Major, P. Schmid, S. Gug- ger, ...

    2021

  10. [18]

    B. W. Matthews , Comparison of the predicted and observed secondary structure of T4 phage lysozyme, Biochimica et Biophysica Acta (BBA) - Protein Struc- ture, 405 (1975), pp. 442–451

    1975

  11. [19]

    McElfresh, S

    D. McElfresh, S. Khandagale, J. V alverde, V. Prasad C, G. Ramakrishnan, M. Goldblum, and C. White , When do neural nets outperform boosted trees on tabular data?, Advances in Neural Information Processing Systems, 36 (2024)

    2024

  12. [20]

    V. Mnih, K. Kavukcuoglu, D. Silver, A. Graves, I. Antonoglou, D. Wierstra, and M. Riedmiller, Playing atari with deep reinforcement learning, 2013

    2013

  13. [21]

    Pandey and A

    P. Pandey and A. Srivastava, samp-vgg16: Drude polarizable force-field assisted image-based deep neural network prediction model for short antimicrobial peptides, bioRxiv, (2023), pp. 2023–06

    2023

  14. [22]

    Park and et al

    I.-B. Park and et al. , A reinforcement learning approach to robust scheduling of semiconductor manufacturing facilities, IEEE Transactions on Au- tomation Science and Engineering, (2019)

    2019

  15. [23]

    Popov, S

    S. Popov, S. Morozov, and A. Babenko , Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data, Sept. 2019. arXiv:1909.06312 [cs, stat]

    2019 arXiv

  16. [24]

    L. O. Prokhorenkova, G. Gusev, A. Vorobev, A. V. Dorogush, and A. Gulin, Catboost: unbiased boosting with categorical features., in NeurIPS, S. Ben- gio, H. M. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett, eds., 2018, pp. 6639– 6649

    2018

  17. [25]

    Rajkomar and et al

    A. Rajkomar and et al. , Machine learning in medicine, New England Journal of Medicine, (2019)

    2019

  18. [26]

    Salem, A

    M. Salem, A. Keshavarzi Arshadi, and J. S. Yuan, Ampdeep: hemolytic activity prediction of antimicrobial peptides using transfer learning, BMC bioinformatics, 23 (2022), p. 389

    2022

  19. [27]

    Sch ¨afl, L

    B. Sch ¨afl, L. Gruber, A. Bitto-Nemling, and S. Hochreiter , Hopular: Modern hopfield networks for tabular data, 2022

    2022

  20. [28]

    Shwartz-Ziv and A

    R. Shwartz-Ziv and A. Armon , Tabular Data: Deep Learning is Not All You Need, Nov. 2021. 561 citations (Semantic Scholar/arXiv) [2024-01-16] arXiv:2106.03253 [cs]

    2021 arXiv

  21. [29]

    Soueidan and M

    H. Soueidan and M. Nikolski , Machine learning Submitted to SIAM Conference on Data Mining for metagenomics: methods and tools, arXiv preprint arXiv:1510.06621, (2015)

    2015 arXiv

  22. [30]

    V anschoren, J

    J. V anschoren, J. N. van Rijn, B. Bischl, and L. Torgo , Openml: networked science in machine learning, ArXiv, abs/1407.7722 (2014)

    2014 arXiv

  23. [31]

    V aswani, N

    A. V aswani, N. Shazeer, N. Parmar, J. Uszko- reit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, Attention Is All You Need, Aug. 2023. arXiv:1706.03762 [cs]

    2023 arXiv

  24. [32]

    Wydma ´nski, O

    W. Wydma ´nski, O. Bulenok, and M. ´Smieja, HyperTab: Hypernetwork Approach for Deep Learning on Small Tabular Datasets, Aug. 2023. arXiv:2304.03543 [cs]

    2023 arXiv

  25. [33]

    Yosinski, J

    J. Yosinski, J. Clune, Y. Bengio, and H. Lipson , How transferable are features in deep neural networks?, Advances in neural information processing systems, 27 (2014)

    2014

  26. [34]

    Young, D

    T. Young, D. Hazarika, S. Poria, and E. Cam- bria, Recent Trends in Deep Learning Based Natural Language Processing, Nov. 2018. arXiv:1708.02709 [cs]

    2018 arXiv

  27. [35]

    0.1g"), PROJ_LR = float(1e-5, 1e-3,

    B. Zhu, X. Shi, N. Erickson, M. Li, G. Karypis, and M. Shoaran , Xtab: Cross-table pretraining for tabular transformers, arXiv preprint arXiv:2305.06090, (2023). A Experimental setup To aid in reproducing the results, we present technical details regarding our experiments. Ini...

    2023 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.