Pith. sign in

REVIEW 3 major objections 5 minor 107 references

Drift-Resilient TabPFN: In-Context Learning Temporal Distribution Shifts on Tabular Data

T0 review · 3 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 A single fixed transformer, pretrained on synthetic causal models with drifting edges, predicts future temporal domains on small tabular datasets better than tuned gradient-boosted trees and the original TabPFN.

desk verdict Genuinely new prior for drift-aware TabPFN with a thorough evaluation, but the 'consistently outperforms' claim outruns the aggregated evidence. read the letter →

arxiv 2411.10634 v1 pith:QOE4UPX7 submitted 2024-11-15 cs.LG stat.ML

Kai Helli , David Schnurr , Noah Hollmann , Samuel Müller , Frank Hutter This is my paper · ORCID
classification cs.LGstat.ML
keywords temporaldistributionshifttabulardatain-contextlearningprior-datafittednetworksstructuralcausalmodelsdomaingeneralizationout-of-distributionpredictioncalibration
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that a fixed, pretrained transformer can learn to handle temporal distribution shifts on small tabular datasets by being trained once on millions of synthetic datasets in which the causal rules generating the data drift over time. Rather than tuning a model to each dataset, the network uses in-context learning and the time index of each sample to extrapolate the decision boundary into future, unseen domains. Across 18 synthetic and real-world datasets, the paper reports that this Drift-Resilient TabPFN outperforms tuned GBDTs, the original TabPFN, and Wild-Time baselines on out-of-distribution data, raising combined accuracy from 0.688 to 0.744 and ROC AUC from 0.786 to 0.832, while producing probability estimates that are better calibrated.

What carries the argument

The load-bearing object is the 2nd-order SCM, a secondary structural causal model whose functional-graph forward pass maps a temporal domain index, encoded with Time2Vec, to weight shifts for a sparse set of edges in the primary SCM that generates each synthetic dataset. Because the secondary SCM has nonlinear mechanisms and the shifted edge weights feed through the primary SCM, the prior produces covariate shifts, prior probability shifts, concept shifts, and combinations thereof, all correlated across features and extrapolating beyond training domains. The transformer (TabPFN) is trained to approximate Bayesian inference over this prior, so at inference it conditions on the training set plus domain indices and outputs posterior predictions.

What would settle it

Run the model on a benchmark dataset whose shift is a sudden, discontinuous regime change (a change point or feature-space alteration) rather than gradual edge-weight drift, using the same Eval-Fix protocol; if Drift-Resilient TabPFN no longer beats a last-domain-only XGBoost or the original TabPFN on OOD accuracy and ECE, the claimed generality of the prior is refuted. The paper's own Appendix A.4.2 already shows that perturbed domain indices degrade performance, so a cleaner test would compare accurate against noisy domain labels on a real-world temporal split.

Watch

Extended reading notes

Core claim

The central claim is that temporal distribution shifts in tabular data can be modeled as sparse, gradual changes to the edge weights of a structural causal model, and that a Prior-Data Fitted Network trained on a prior made of such drifting SCMs will learn to detect and extrapolate these shifts on unseen data. The paper introduces a secondary, '2nd-order' SCM that takes a temporal domain index as input and outputs correlated shift parameters for selected edges in the primary SCM's functional graph. After pretraining on 30.72 million synthetic datasets, the resulting model, applied in one forward pass without hyperparameter tuning, is claimed to consistently beat all baselines on OOD data across 18 benchmarks while maintaining stronger calibration than methods that ignore shifts.

Load-bearing premise

The load-bearing premise is that real-world temporal shifts are well approximated by sparse, gradual, correlated changes to the causal edge weights of a structural causal model, driven by a secondary process indexed by time, with accurate temporal domain labels available at inference.

Editorial extensions

If this is right

  • A single fixed model, with no per-dataset tuning, extrapolates to future temporal domains better than tuned GBDTs on small and moderate tabular datasets.
  • The same prior covers covariate shift, prior probability shift, concept shift, and their combinations, so users do not need to diagnose the shift type in advance.
  • Out-of-distribution probability estimates are better calibrated than those of baselines, lowering expected calibration error from 0.119 to 0.091 on the combined benchmark.
  • Training and prediction together take on average 10.9 seconds, making drift handling practical where retraining pipelines are expensive.

Reading between the lines

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

  • The paper's prior assumes smooth edge-weight drift; the likely failure mode is abrupt regime changes or unobserved confounders, which the prior cannot express, so a natural extension is to add change-point processes to the 2nd-order SCM and test on datasets with sudden shifts.
  • Because the pretraining prior, not Time2Vec, carries most of the gain (ablation in Table 2), the same drift prior could be dropped into later TabPFN variants, such as dataset-distillation or prompt-tuning versions, to extend the approach to larger datasets.
  • The sensitivity to domain-index quality (Appendix A.4.2) suggests the practical ceiling on real-world data may be set by how well domain indices can be approximated from features, so learning the domain index end-to-end would be a direct test.
  • If domain indices are themselves uncertain, an online or transductive variant that updates beliefs about the current domain could recover some of the lost performance; the paper flags these settings as future work.
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

3 major / 5 minor

Summary. The paper proposes Drift-Resilient TabPFN, an extension of the TabPFN framework for tabular classification under temporal distribution shifts. The main idea is to modify the synthetic-data prior used to pre-train the transformer: a structural causal model (SCM) is sampled, a sparse subset of its edges is selected, and a second-order SCM generates smooth, correlated changes to those edge weights as a function of a temporal domain index encoded with Time2Vec. The model is pre-trained once on datasets drawn from this drift prior and is then applied without per-dataset tuning. The authors evaluate the method on 18 datasets (8 synthetic, 10 real-world), comparing against GBDTs, the original TabPFN, and Wild-Time methods, and report aggregate OOD accuracy, F1, ROC AUC, and ECE with 95% confidence intervals. They claim consistent OOD improvements and stronger calibration, with combined accuracy improving from 0.688 to 0.744 and ROC AUC from 0.786 to 0.832 relative to the strongest baselines.

Significance. If the central claim holds, this is a meaningful contribution to temporal domain generalization on small tabular data: it is, to my knowledge, the first tabular method to show consistent OOD gains over shift-agnostic supervised learning in this setting, and it does so with a single fixed model that requires no per-dataset hyperparameter tuning. The prior construction is novel and unifies covariate, prior-probability, and concept shift within one generative framework. The evaluation is thorough in several respects: 18 datasets, multiple splits and seeds, 95% confidence intervals, HPO saturation checks, a Time2Vec ablation, and a decision-boundary analysis. The authors also release code, pre-trained models, and a Colab notebook, and the model is pre-trained without access to real test labels, so the central empirical claim is not a fitted prediction. The main weakness is that the 'consistent outperformance' claim is supported only by aggregate means; per-dataset results are not reported, and the real-world ROC AUC gains are small and statistically overlapping with those of the original TabPFN.

major comments (3)
  1. [Section 4, Tables 1 and 5, Abstract] The central claim that Drift-Resilient TabPFN 'consistently outperforms all baselines' on OOD data is not supported by the evidence as presented. Tables 1 and 5 report only aggregated means over 8 synthetic and 10 real-world datasets; no per-dataset results are shown anywhere in the paper or appendix. The 95% confidence intervals are computed across model initializations (and after averaging over three splits), so they do not quantify dataset-level variability. A model can have a higher mean ROC AUC while losing on a substantial fraction of datasets. Since the synthetic gains are large (Table 1: OOD ROC 0.749 to 0.844) and the real-world gains are much smaller (Table 1: 0.820 to 0.822), the aggregate could be dominated by the synthetic subset. To establish the 'consistent' component of the claim, please provide a per-dataset table (or supplement) listing OOD accuracy and ROC AUC for every dataset and every main method, together with the number and percentage of datasets where Drift-Resilient TabPFN wins, ties, or loses against each baseline, and a paired statistical test over datasets (e.g., Wilcoxon signed-rank) using the per-dataset scores. This is directly checkable given the released code.
  2. [Table 1, real-world block; Abstract] The real-world OOD ROC AUC improvement is negligible: 0.822 ± 0.010 for Drift-Resilient TabPFN versus 0.816 ± 0.006 for TabPFNbase with domain indices, with clearly overlapping confidence intervals. The combined ROC AUC gain from 0.786 to 0.832 highlighted in the abstract is driven primarily by the synthetic subset (0.749 to 0.844). The paper should either present the real-world and synthetic results separately in the main claims or provide per-dataset evidence that the real-world ROC AUC difference is not noise. As written, the abstract's 'large performance improvements' overstates the real-world evidence, especially for ROC AUC.
  3. [Section 4, Evaluation Strategy; Appendix A.4.4.2] The statistical reporting is ambiguous and too weak for the dataset-level claim. The text says 'We generate three random splits and average metrics across these splits. Each method is trained three times, and we report the average and 95%-confidence intervals calculated across model initializations.' This does not describe a valid procedure for obtaining confidence intervals for the mean over datasets; if the intervals are over three initializations, they ignore between-dataset variance. Please clarify exactly how the confidence intervals were computed (e.g., bootstrap over dataset means, or over runs within a fixed dataset) and, if the claim is about datasets, report dataset-level variability. The critical difference diagrams in Figure 10 are also uninterpretable without the underlying per-dataset rank table; please include that table or state where it can be found.
minor comments (5)
  1. [Section 3.2, Eq. (1)] Please clarify the status of the Time2Vec parameters ω_i and φ_i: are they learned during pre-training only, or also adapted at inference, and does the reported inference time of 10.9s include any Time2Vec fitting?
  2. [Table 2 and Section 5] The Time2Vec ablation shows that the contribution of Time2Vec is statistically insignificant, yet it is kept because HPO favored it. This is a reasonable model-selection decision, but the paper should state explicitly that the reported gains should not be attributed to Time2Vec, to avoid reader confusion.
  3. [Figures 5, 7, 8, and 9] The decision-boundary figures are very small and the differences between rows are hard to see in print. Please enlarge them or split them across pages, and make sure the color scheme is accessible to color-blind readers.
  4. [Tables 1, 5, and 6] The tables are dense and use many repeated model names; consider grouping rows by method family or using horizontal rules to make the comparisons easier to follow. Also ensure the bold 'best mean' marking is applied consistently across all metric columns, including ECE where lower is better.
  5. [Appendix A.4.3.4] The comparison against DRAIN and GI on Rotated Two Moons reuses figures from the DRAIN paper; please confirm that reproduction permission is covered by the paper's license or add a note in the caption.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the fixed pretrained model is evaluated on external benchmarks, and no prediction reduces to a fitted parameter or self-citation chain.

full rationale

The paper's derivation chain is: define an SCM-based prior with temporal edge shifts generated by a secondary SCM (Section 3.2, Algorithm 1), pretrain a Prior-Data Fitted Network on millions of datasets sampled from this prior (Section 3.1), then apply the fixed pretrained model to unseen synthetic and real-world datasets in a single forward pass (Section 4). No model parameter is fitted to the test labels, and the pretrained model is not adapted per dataset; the central empirical claim is therefore not a fitted prediction renamed as a prediction. The synthetic test datasets described in Appendix A.7.2.1 are external generators (scikit-multiflow and custom geometric datasets) rather than fresh samples from the pretraining prior, so the evaluation is not tautological. The self-citations to Müller et al. [18] and Hollmann et al. [19] supply the PFN/TabPFN mechanism, but the paper's contribution is the new temporal-shift prior and its empirical validation, which stands on independent benchmarks; no load-bearing argument reduces to those citations. The limitations section explicitly concedes that the sparse-mechanism-shift prior may not capture all real-world shifts, which is a scope limitation rather than circularity. The lack of per-dataset performance breakdowns is an evidentiary gap about the word 'consistently,' not a circularity. Accordingly, no circular step can be exhibited by quoting an equation or a fitted parameter that is equivalent to the claimed output.

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

The central claim rests on the SCM-based prior and its shift dynamics. The 2nd-order SCM H and the sparse edge-shift mechanism are introduced by this paper without independent evidence. Preprocessing hyperparameters were tuned on validation data. The exact prior distributions are not fully specified, so the free-parameter count is likely underreported.

free parameters (4)
  • Sparse edge shift selection and shift magnitude distributions in the prior
    The prior samples a sparse subset of SCM edges to shift and the 2nd-order SCM generates shift magnitudes. The exact distributions and hyperparameters are not specified in the main text and were partly tuned on validation.
  • Number of functional subnodes k and l in SCM expansion
    These control the size and complexity of the generated functional graphs but are not reported.
  • Preprocessing hyperparameters (softmax_temperature, use_poly_features, remove_outliers, etc.) = optimized on validation, see Table 10
    Selected by random search over 300 configurations on 12 validation datasets to maximize OOD ROC AUC.
  • Time2Vec dimension m and learned parameters
    m-dimensional temporal encoding with learned weights; decision to keep Time2Vec was guided by HPO (Appendix A.4.1).
assumptions (4)
  • domain assumption Real-world temporal distribution shifts can be modeled as gradual changes to the edge weights of an SCM.
    Stated in Section 3.2 and acknowledged as a limitation in Section 5: 'The underlying prior for structural causal models with sparse mechanism shifts may not accurately describe all real-world datasets.'
  • standard math PFN training on synthetic datasets approximates Bayesian inference for the specified prior.
    Borrowed from Müller et al. [18]; the paper relies on this to justify the model's behavior.
  • domain assumption Ground-truth temporal domain indices are available for training.
    Appendix A.4.2 shows perturbing domain indices degrades performance; real-world data only provides approximations.
  • domain assumption The SCM prior is a sufficiently general approximation to many real-world tabular datasets.
    Inherited from TabPFN [19]; used to justify the SCM-based synthetic data generator.
invented entities (1)
  • 2nd-order SCM H
    purpose: Generates edge shifts for the primary SCM based on temporal domain input, enabling correlated, nonlinear, extrapolating shift dynamics.
    H is a heuristic generative model introduced by this paper. It has no falsifiable handle outside the paper; its validity is only assessed indirectly via downstream classification performance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Drift-Resilient TabPFN: In-Context Learning Temporal Distribution Shifts on Tabular Data." pith.science (2026). https://pith.science/paper/QOE4UPX7

@misc{pith2026241110634,
  author       = {Pith},
  title        = {Pith review of: Drift-Resilient TabPFN: In-Context Learning Temporal Distribution Shifts on Tabular Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QOE4UPX7}},
  note         = {Machine review of arXiv:2411.10634}
}
read the original abstract

While most ML models expect independent and identically distributed data, this assumption is often violated in real-world scenarios due to distribution shifts, resulting in the degradation of machine learning model performance. Until now, no tabular method has consistently outperformed classical supervised learning, which ignores these shifts. To address temporal distribution shifts, we present Drift-Resilient TabPFN, a fresh approach based on In-Context Learning with a Prior-Data Fitted Network that learns the learning algorithm itself: it accepts the entire training dataset as input and makes predictions on the test set in a single forward pass. Specifically, it learns to approximate Bayesian inference on synthetic datasets drawn from a prior that specifies the model's inductive bias. This prior is based on structural causal models (SCM), which gradually shift over time. To model shifts of these causal models, we use a secondary SCM, that specifies changes in the primary model parameters. The resulting Drift-Resilient TabPFN can be applied to unseen data, runs in seconds on small to moderately sized datasets and needs no hyperparameter tuning. Comprehensive evaluations across 18 synthetic and real-world datasets demonstrate large performance improvements over a wide range of baselines, such as XGB, CatBoost, TabPFN, and applicable methods featured in the Wild-Time benchmark. Compared to the strongest baselines, it improves accuracy from 0.688 to 0.744 and ROC AUC from 0.786 to 0.832 while maintaining stronger calibration. This approach could serve as significant groundwork for further research on out-of-distribution prediction.

Figures

Figures reproduced from arXiv: 2411.10634 by the authors.

Figure 1
Figure 1. High-level overview of our method. We train a transformer that accepts entire datasets as [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Illustrative transformation of an SCM to one exemplary functional representation. Shaded [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Diagram illustrating the integration of a [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Types of distribution shifts based on the definitions by Moreno-Torres et al. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: This figure displays the predictive behavior of TabPFN [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: This figure shows the temporal shifts of two synthetic datasets across selected domains. [PITH_FULL_IMAGE:figures/full_fig_p020_6.png]
Figure 7
Figure 7. Figure 7: This figure contrasts the predictive behavior of TabPFN [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]
Figure 8
Figure 8. Figure 8: Each figure displays the predictive behavior of TabPFN [PITH_FULL_IMAGE:figures/full_fig_p021_8.png]
Figure 9
Figure 9. Figure 9: Comparison of our method against DRAIN [ [PITH_FULL_IMAGE:figures/full_fig_p021_9.png]
Figure 10
Figure 10. Figure 10: This figure presents critical difference diagrams of our evaluated metrics on OOD [PITH_FULL_IMAGE:figures/full_fig_p023_10.png]
Figure 11
Figure 11. Figure 11: This figure illustrates a comparative analysis of the resilience of the listed methods to [PITH_FULL_IMAGE:figures/full_fig_p026_11.png]
Figure 12
Figure 12. Figure 12: This figure shows the performance of Drift-Resilient TabPFN and the baseline TabPFN on [PITH_FULL_IMAGE:figures/full_fig_p027_12.png]
Figure 13
Figure 13. Figure 13: Illustration of initial and shifted data distributions alongside their optimal decision [PITH_FULL_IMAGE:figures/full_fig_p028_13.png]
Figure 14
Figure 14. Figure 14: Share of temporal domains in exemplary datasets prior seen up to any instance [PITH_FULL_IMAGE:figures/full_fig_p028_14.png]
Figure 15
Figure 15. Figure 15: This figure presents three exemplary functions sampled from nodes within the network of [PITH_FULL_IMAGE:figures/full_fig_p028_15.png]
Figure 16
Figure 16. Figure 16: Adapted from the Wild-Time benchmark [11], this illustration portrays the Eval-Fix evaluation strategy employed in our study. The domain boundary is indicated by ct, beyond which datasets are considered part of the out-of-distribution (OOD) test set DOOD. To evaluate …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

107 extracted references · 57 canonical work pages

  1. [1]

    Causal discovery in heterogeneous environments under the sparse mechanism shift hypothesis

    Ronan Perry, Julius V on Kügelgen, and Bernhard Schölkopf. Causal discovery in heterogeneous environments under the sparse mechanism shift hypothesis. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho, editors, Proceedings of the 36th International Con- ference on Advances in Neural Information Processing Systems (NeurIPS’22) , 2022. URL...

    2022

  2. [2]

    Daniel Vela, Andrew Sharp, Richard Zhang, Trang Nguyen, An Hoang, and Oleg S. Pianykh. Temporal quality degradation in AI models. Scientific Reports, 12(1):11654, July 2022. ISSN 2045-2322. doi:10.1038/s41598-022-15245-z

    doi:10.1038/s41598-022-15245-z 2022

  3. [3]

    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

    2022

  4. [4]

    Why tabular foundation models should be a research priority

    Boris van Breugel and Mihaela van der Schaar. Why tabular foundation models should be a research priority. arXiv preprint arXiv:2405.01147, 2024

    arXiv 2024

  5. [5]

    Empirical evaluation of performance degradation of machine learning-based predictive models–a case study in healthcare information systems

    Zachary Young and Robert Steele. Empirical evaluation of performance degradation of machine learning-based predictive models–a case study in healthcare information systems. International Journal of Information Management Data Insights, 2(1):100070, 2022

    2022

  6. [6]

    Temporal shifts in clinical presen- tation and underlying mechanisms of atherosclerotic disease

    Gerard Pasterkamp, Hester M Den Ruijter, and Peter Libby. Temporal shifts in clinical presen- tation and underlying mechanisms of atherosclerotic disease. Nature Reviews Cardiology, 14 (1):21–29, 2017

    2017

  7. [7]

    Mortality prediction of covid-19 patients at intensive care unit admission

    Rajarajan Ganesan, Varun Mahajan, Karan Singla, Sushant Konar, Tanvir Samra, Senthil K Sundaram, Vikas Suri, Mandeep Garg, Naveen Kalra, Goverdhan D Puri, et al. Mortality prediction of covid-19 patients at intensive care unit admission. Cureus, 13(11), 2021

    2021

  8. [8]

    Climate-invariant machine learning

    Tom Beucler, Pierre Gentine, Janni Yuval, Ankitesh Gupta, Liran Peng, Jerry Lin, Sungduk Yu, Stephan Rasp, Fiaz Ahmed, Paul A O’Gorman, et al. Climate-invariant machine learning. Science Advances, 10(6):eadj7250, 2024

    2024

Show all 107 references
  1. [9]

    Dataset shift quantification for credit card fraud detection

    Yvan Lucas, Pierre-Edouard Portier, Léa Laporte, Sylvie Calabretto, Liyun He-Guelton, Frederic Oble, and Michael Granitzer. Dataset shift quantification for credit card fraud detection. In2019 IEEE second international conference on artificial intelligence and knowledge engine...

    2019

  2. [10]

    Hidden risks of machine learning applied to healthcare: Unintended feedback loops between models and future data causing model degradation

    George Alexandru Adam, Chun-Hao Kingsley Chang, Benjamin Haibe-Kains, and Anna Goldenberg. Hidden risks of machine learning applied to healthcare: Unintended feedback loops between models and future data causing model degradation. In Finale Doshi-Velez, Jim Fackler, Ken Jung, ...

    2020

  3. [11]

    Wild-time: A benchmark of in-the-wild distribution shift over time

    Huaxiu Yao, Caroline Choi, Bochuan Cao, Yoonho Lee, Pang Wei Koh, and Chelsea Finn. Wild-time: A benchmark of in-the-wild distribution shift over time. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho, editors,Proceedings of the 36th International Conference...

    2022

  4. [12]

    Temporal domain generalization with drift-aware dynamic neural networks

    Guangji Bai, Chen Ling, and Liang Zhao. Temporal domain generalization with drift-aware dynamic neural networks. In Proceedings of the International Conference on Learning Rep- resentations (ICLR’23), 2023. URL https://openreview.net/forum?id=sWOsRj4nT1n. Published online: iclr.cc

    2023

  5. [13]

    Training for the future: A simple gradient interpolation loss to generalize along time

    Anshul Nasery, Soumyadeep Thakur, Vihari Piratla, Abir De, and Sunita Sarawagi. Training for the future: A simple gradient interpolation loss to generalize along time. In M. Ranzato, A. Beygelzimer, K. Nguyen, P. Liang, J. Vaughan, and Y . Dauphin, editors, Proceedings of the ...

    2021

  6. [14]

    Benchmarking distribution shift in tabular data with tableshift

    Josh Gardner, Zoran Popovic, and Ludwig Schmidt. Benchmarking distribution shift in tabular data with tableshift. Advances in Neural Information Processing Systems, 36, 2024

    2024

  7. [15]

    Revisiting deep learning models for tabular data

    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

    2021

  8. [16]

    Tabular data: Deep learning is not all you need

    Ravid Shwartz-Ziv and Amitai Armon. Tabular data: Deep learning is not all you need. Information Fusion, 81:84–90, 2022

    2022

  9. [17]

    Why do tree-based models still outperform deep learning on typical tabular data? In Alice H

    Leo Grinsztajn, Edouard Oyallon, and Gael Varoquaux. Why do tree-based models still outperform deep learning on typical tabular data? In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho, editors, Proceedings of the 36th International Conference on Advances in N...

    2022

  10. [18]

    Müller, N

    S. Müller, N. Hollmann, S. Arango, J. Grabocka, and F. Hutter. Transformers can do bayesian inference. In Proceedings of the International Conference on Learning Representations (ICLR’22), 2022. URL https://openreview.net/forum?id=KSugKcbNf9. Published online: iclr.cc

    2022

  11. [19]

    Hollmann, S

    N. Hollmann, S. Müller, K. Eggensperger, and F. Hutter. TabPFN: A transformer that solves small tabular classification problems in a second. InProceedings of the International Conference on Learning Representations (ICLR’23), 2023. URL https://openreview.net/forum?id= cp5PvcI6...

    2023

  12. [20]

    Causality

    Judea Pearl. Causality. Cambridge University Press, 2 edition, 2009

    2009

  13. [21]

    Peters, D

    J. Peters, D. Janzing, and B. Schölkopf. Elements of causal inference: foundations and learning algorithms. The MIT Press, 2017

    2017

  14. [22]

    Time2vec: Learning a vector representation of time, 2020

    Seyed Mehran Kazemi, Rishab Goel, Sepehr Eghbali, Janahan Ramanan, Jaspreet Sahota, Sanjay Thakur, Stella Wu, Cathal Smyth, Pascal Poupart, and Marcus Brubaker. Time2vec: Learning a vector representation of time, 2020. URL https://openreview.net/forum? id=rklklCVYvB

    2020

  15. [23]

    Chen and C

    T. Chen and C. Guestrin. Xgboost: A scalable tree boosting system. In B. Krishnapuram, M. Shah, A. Smola, C. Aggarwal, D. Shen, and R. Rastogi, editors, Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD’16), pages 785–794. ...

    2016

  16. [24]

    Prokhorenkova, G

    L. Prokhorenkova, G. Gusev, A. V orobev, A. Dorogush, and A. Gulin. CatBoost: unbiased boosting with categorical features. In S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett, editors, Proceedings of the 31st International Conference on Advance...

    2018

  17. [25]

    G. Ke, Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye, and T.-Y . Liu. Lightgbm: A highly efficient gradient boosting decision tree. In I. Guyon, U. von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, editors, Proceedings of the 30th Internation...

    2017

  18. [26]

    Continuously indexed domain adaptation

    Hao Wang, Hao He, and Dina Katabi. Continuously indexed domain adaptation. In H. Daume III and A. Singh, editors, Proceedings of the 37th International Conference on Machine Learning (ICML’20), volume 98. Proceedings of Machine Learning Research, 2020

    2020

  19. [27]

    Selçuk Candan, Adrienne Raglin, and Huan Liu

    Paras Sheth, Raha Moraffah, K. Selçuk Candan, Adrienne Raglin, and Huan Liu. Domain generalization – a causal perspective, 2022

    2022

  20. [28]

    Domain generalization via invariant feature representation

    Krikamol Muandet, David Balduzzi, and Bernhard Schölkopf. Domain generalization via invariant feature representation. In S. Dasgupta and D. McAllester, editors, Proceedings of the 30th International Conference on Machine Learning (ICML’13). Omnipress, 2013. URL https://proceed...

    2013

  21. [29]

    Metareg: Towards domain generalization using meta-regularization

    Yogesh Balaji, Swami Sankaranarayanan, and Rama Chellappa. Metareg: Towards domain generalization using meta-regularization. In S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett, editors, Proceedings of the 31st International Conference on Advan...

    2018

  22. [30]

    Unified deep supervised domain adaptation and generalization

    Saeid Motiian, Marco Piccirilli, Donald A Adjeroh, and Gianfranco Doretto. Unified deep supervised domain adaptation and generalization. In Proceedings of the IEEE international conference on computer vision, pages 5715–5725, 2017

    2017

  23. [31]

    Invariant risk mini- mization, 2020

    Martin Arjovsky, Léon Bottou, Ishaan Gulrajani, and David Lopez-Paz. Invariant risk mini- mization, 2020

    2020

  24. [32]

    Hashimoto, and Percy Liang

    Shiori Sagawa, Pang Wei Koh, Tatsunori B. Hashimoto, and Percy Liang. Distributionally robust neural networks for group shifts: On the importance of regularization for worst-case generalization. In Proceedings of the International Conference on Learning Representations (ICLR’2...

    2020

  25. [33]

    Dauphin, and David Lopez-Paz

    Hongyi Zhang, Moustapha Cisse, Yann N. Dauphin, and David Lopez-Paz. mixup: Beyond empirical risk minimization. In Proceedings of the International Conference on Learning Representations (ICLR’18), 2018. Published online: iclr.cc

    2018

  26. [34]

    Improving out-of-distribution robustness via selective augmentation, 2022

    Huaxiu Yao, Yu Wang, Sai Li, Linjun Zhang, Weixin Liang, James Zou, and Chelsea Finn. Improving out-of-distribution robustness via selective augmentation, 2022

    2022

  27. [35]

    Deep coral: Correlation alignment for deep domain adaptation

    Baochen Sun and Kate Saenko. Deep coral: Correlation alignment for deep domain adaptation. In Gang Hua and Hervé Jégou, editors, Computer Vision – ECCV 2016 Workshops , pages 443–450, Cham, 2016. Springer International Publishing. doi:10.1007/978-3-319-49409-8_35

    2016 doi

  28. [36]

    Domain-adversarial training of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pascal Germain, Hugo Larochelle, François Laviolette, Mario Marchand, and Victor Lempitsky. Domain-adversarial training of neural networks. Journal of Machine Learning Research, 17(1):2096–2030, jan 2016. ISSN 1532- 4435

    2016

  29. [37]

    Out-of-distribution generalization via risk extrapo- lation (rex)

    David Krueger, Ethan Caballero, Joern-Henrik Jacobsen, Amy Zhang, Jonathan Binas, Dinghuai Zhang, Remi Le Priol, and Aaron Courville. Out-of-distribution generalization via risk extrapo- lation (rex). In M. Meila and T. Zhang, editors,Proceedings of the 38th International Conf...

    2021

  30. [38]

    Pappas, and Bernhard Schölkopf

    Cian Eastwood, Alexander Robey, Shashank Singh, Julius von Kügelgen, Hamed Hassani, George J. Pappas, and Bernhard Schölkopf. Probable domain generalization via quantile risk minimization. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho, editors, Proceeding...

    2022

  31. [39]

    Andrey Malinin, Neil Band, Alexander Ganshin, German Chesnokov, Yarin Gal, Mark J. F. Gales, Alexey Noskov, Andrey Ploskonosov, Liudmila Prokhorenkova, Ivan Provilkov, Vat- sal Raina, Vyas Raina, Denis Roginskiy, Mariya Shmatova, Panos Tigar, and Boris Yangel. Shifts: A datase...

    2021 arXiv

  32. [40]

    Andrey Malinin, Andreas Athanasopoulos, Muhamed Barakovic, Meritxell Bach Cuadra, Mark J. F. Gales, Cristina Granziera, Mara Graziani, Nikolay Kartashev, Konstantinos Kyriakopoulos, Po-Jui Lu, Nataliia Molchanova, Antonis Nikitakis, Vatsal Raina, Francesco La Rosa, Eli Sivena,...

    2022 arXiv

  33. [41]

    On the need for a language describing distribution shifts: Illustrations on tabular datasets

    Jiashuo Liu, Tianyu Wang, Peng Cui, and Hongseok Namkoong. On the need for a language describing distribution shifts: Illustrations on tabular datasets. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2023

    2023

  34. [42]

    Retiring adult: New datasets for fair machine learning

    Frances Ding, Moritz Hardt, John Miller, and Ludwig Schmidt. Retiring adult: New datasets for fair machine learning. In M. Ranzato, A. Beygelzimer, K. Nguyen, P. Liang, J. Vaughan, and Y . Dauphin, editors,Proceedings of the 34th International Conference on Advances in Neural ...

    2021

  35. [43]

    Wild-tab: A benchmark for out-of-distribution generalization in tabular regression, 2023

    Sergey Kolesnikov. Wild-tab: A benchmark for out-of-distribution generalization in tabular regression, 2023. URL https://arxiv.org/abs/2312.01792

    2023 arXiv

  36. [44]

    In search of lost domain generalization

    Ishaan Gulrajani and David Lopez-Paz. In search of lost domain generalization. In Proceedings of the International Conference on Learning Representations (ICLR’21), 2021. URL https: //openreview.net/forum?id=lQdXeXDoWtI. Published online: iclr.cc

    2021

  37. [45]

    Jindong Wang, Cuiling Lan, Chang Liu, Yidong Ouyang, Tao Qin, Wang Lu, Yiqiang Chen, Wenjun Zeng, and Philip S. Yu. Generalizing to unseen domains: A survey on domain generalization. IEEE Transactions on Knowledge and Data Engineering, 35(8):8052–8072,

  38. [46]

    Averaging weights leads to wider optima and better generalization

    Pavel Izmailov, Dmitrii Podoprikhin, Timur Garipov, Dmitry Vetrov, and Andrew Gordon Wilson. Averaging weights leads to wider optima and better generalization. In A. Globerson and R. Silva, editors, Proceedings of The 34th Uncertainty in Artificial Intelligence Conference (UAI...

    2018

  39. [47]

    Task agnostic continual learning via meta learning

    Xu He, Jakub Sygnowski, Alexandre Galashov, Andrei Alex Rusu, Yee Whye Teh, and Razvan Pascanu. Task agnostic continual learning via meta learning. In 4th Lifelong Machine Learning Workshop at ICML 2020, 2020. URL https://openreview.net/forum?id=AeIzVxdJgeb

    2020

  40. [48]

    In-context data distillation with tabpfn, 2024

    Junwei Ma, Valentin Thomas, Guangwei Yu, and Anthony Caterini. In-context data distillation with tabpfn, 2024. URL https://arxiv.org/abs/2402.06971

    2024 arXiv

  41. [49]

    Tunetables: Context optimization for scalable prior-data fitted networks

    Benjamin Feuer, Robin Tibor Schirrmeister, Valeriia Cherepanova, Chinmay Hegde, Frank Hutter, Micah Goldblum, Niv Cohen, and Colin White. Tunetables: Context optimization for scalable prior-data fitted networks. arXiv preprint arXiv:2402.11137, 2024

    2024 arXiv

  42. [50]

    Forecastpfn: Synthetically-trained zero-shot forecasting

    Samuel Dooley, Gurnoor Singh Khurana, Chirag Mohapatra, Siddartha V Naidu, and Colin White. Forecastpfn: Synthetically-trained zero-shot forecasting. In Advances in Neural Information Processing Systems, 2023

    2023

  43. [51]

    Moreno-Torres, Troy Raeder, Rocío Alaiz-Rodríguez, Nitesh V

    Jose G. Moreno-Torres, Troy Raeder, Rocío Alaiz-Rodríguez, Nitesh V . Chawla, and Fran- cisco Herrera. A unifying view on dataset shift in classification. Pattern Recognition, 45 (1):521–530, 2012. ISSN 0031-3203. doi:10.1016/j.patcog.2011.06.019. URL https: //www.sciencedirec...

    2012 doi

  44. [52]

    Patterns of dataset shift

    Meelis Kull and Peter Flach. Patterns of dataset shift. In First international workshop on learning over multiple contexts (LMCE) at ECML-PKDD, volume 5, 2014

    2014

  45. [53]

    Vanschoren, J

    J. Vanschoren, J. van Rijn, B. Bischl, and L. Torgo. OpenML: Networked science in machine learning. SIGKDD Explorations, 15(2):49–60, 2014. 14

    2014

  46. [54]

    Individual comparisons by ranking methods

    Frank Wilcoxon. Individual comparisons by ranking methods. Biometrics Bulletin, 1(6):80–83,

  47. [55]

    A simple sequentially rejective multiple test procedure

    Sture Holm. A simple sequentially rejective multiple test procedure. Scandinavian journal of statistics, pages 65–70, 1979

    1979

  48. [56]

    statistical comparisons of classifiers over multiple data sets

    Salvador Garcia and Francisco Herrera. An extension on" statistical comparisons of classifiers over multiple data sets" for all pairwise comparisons. Journal of machine learning research, 9 (12), 2008

    2008

  49. [57]

    Deep learning for time series classification: a review.Data Mining and Knowledge Discovery, 33(4):917–963, 2019

    Hassan Ismail Fawaz, Germain Forestier, Jonathan Weber, Lhassane Idoumghar, and Pierre- Alain Muller. Deep learning for time series classification: a review.Data Mining and Knowledge Discovery, 33(4):917–963, 2019

    2019

  50. [58]

    Overcoming catastrophic forgetting in neural networks

    James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska-Barwinska, et al. Overcoming catastrophic forgetting in neural networks. Proceedings of the national academy of scien...

    2017

  51. [59]

    Continual learning through synaptic intelligence

    Friedemann Zenke, Ben Poole, and Surya Ganguli. Continual learning through synaptic intelligence. In D. Precup and Y . Teh, editors,Proceedings of the 34th International Conference on Machine Learning (ICML’17) , volume 70. Proceedings of Machine Learning Research, 2017

    2017

  52. [60]

    Efficient lifelong learning with A-GEM

    Arslan Chaudhry, Marc’Aurelio Ranzato, Marcus Rohrbach, and Mohamed Elhoseiny. Efficient lifelong learning with A-GEM. In Proceedings of the International Conference on Learning Representations (ICLR’19), 2019. Published online: iclr.cc

    2019

  53. [61]

    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 H. Daume III and A. Singh, editors, Proceedings of the 37th International Conference on Machine Learning (ICML’20), volume 98. Proceedin...

    2020

  54. [62]

    Unsupervised learning of visual features by contrasting cluster assignments

    Mathilde Caron, Ishan Misra, Julien Mairal, Priya Goyal, Piotr Bojanowski, and Armand Joulin. Unsupervised learning of visual features by contrasting cluster assignments. In H. Larochelle, M. Ranzato, R. Hadsell, M.-F. Balcan, and H. Lin, editors,Proceedings of the 33rd Intern...

    2020

  55. [63]

    Venkateswarlu

    Bendi Ramana and N. Venkateswarlu. ILPD (Indian Liver Patient Dataset). UCI Machine Learning Repository, 2012

    2012

  56. [64]

    Istanbul Stock Exchange

    Oguz Akbilgic. Istanbul Stock Exchange. UCI Machine Learning Repository, 2013

    2013

  57. [65]

    DeShazo, Chris Gennings, Juan L

    Beata Strack, Jonathan P. DeShazo, Chris Gennings, Juan L. Olmo, Sebastian Ventura, Krzysztof J. Cios, and John N. Clore. Impact of HbA1c Measurement on Hospital Read- mission Rates: Analysis of 70,000 Clinical Database Patient Records. BioMed Research International, 2014:7816...

    2014 doi

  58. [66]

    Data Expo competition

    Albert Bifet and Elena Ikonomovska. Data Expo competition. OpenML, 2009. URL https: //www.openml.org/search?type=data&sort=runs&id=1169&status=active

    2009

  59. [67]

    Everhart, W

    Jack Smith, J. Everhart, W. Dickson, W. Knowler, and Richard Johannes. Using the adap learning algorithm to forcast the onset of diabetes mellitus. Proceedings - Annual Symposium on Computer Applications in Medical Care, 10, 11 1988

    1988

  60. [68]

    M. M. Faniqul Islam, Rahatara Ferdousi, Sadikur Rahman, and Humayra Yasmin Bushra. Likelihood prediction of diabetes at early stage using data mining techniques. In Mousumi Gupta, Debanjan Konar, Siddhartha Bhattacharyya, and Sambhunath Biswas, editors, Computer Vision and Mac...

    2020 doi

  61. [69]

    Occupancy Detection

    Luis Candanedo. Occupancy Detection. UCI Machine Learning Repository, 2016. 15

    2016

  62. [70]

    Behavior of the urban traffic of the city of Sao Paulo in Brazil

    Ricardo Ferreira, Andrea Martiniano, and Renato Sassi. Behavior of the urban traffic of the city of Sao Paulo in Brazil. UCI Machine Learning Repository, 2018

    2018

  63. [71]

    Scikit-multiflow: A multi- output streaming framework

    Jacob Montiel, Jesse Read, Albert Bifet, and Talel Abdessalem. Scikit-multiflow: A multi- output streaming framework. Journal of Machine Learning Research, 19(72):1–5, 2018. URL http://jmlr.org/papers/v19/18-251.html

    2018

  64. [72]

    Use of nonclonal serum immunoglobulin free light chains to predict overall survival in the general population

    Angela Dispenzieri, Jerry Katzmann, Robert Kyle, Dirk Larson, Terry Therneau, Colin Colby, Raynell Clark, Graham Mead, Shaji Kumar, L Melton, and S Rajkumar. Use of nonclonal serum immunoglobulin free light chains to predict overall survival in the general population. Mayo Cli...

    2012 doi

  65. [73]

    Prevalence of monoclonal gammopathy of undetermined significance

    Robert Kyle, Terry Therneau, S Rajkumar, Dirk Larson, Matthew Plevak, Janice Offord, Angela Dispenzieri, Jerry Katzmann, and L Melton. Prevalence of monoclonal gammopathy of undetermined significance. The New England journal of medicine, 354:1362–9, 04 2006. doi:10.1056/NEJMoa054494

    2006 doi

  66. [74]

    Splice-2 comparative evaluation: Electricity Pricing

    Michael Harries. Splice-2 comparative evaluation: Electricity Pricing . University of New South Wales, School of Computer Science and Engineering [Sydney], 1999. URL http: //nla.gov.au/nla.arc-32869

    1999

  67. [75]

    Learning with drift detection

    João Gama, Pedro Medas, Gladys Castillo, and Pedro Rodrigues. Learning with drift detection. In Ana L. C. Bazzan and Sofiane Labidi, editors, Advances in Artificial Intelligence – SBIA 2004, pages 286–295, Berlin, Heidelberg, 2004. Springer Berlin Heidelberg. ISBN 978-3-540- 2...

    2004 doi

  68. [76]

    Absenteeism at work

    Andrea Martiniano and Ricardo Ferreira. Absenteeism at work. UCI Machine Learning Repository, 2018

    2018

  69. [77]

    Heart Disease

    Andras Janosi, William Steinbrunn, Matthias Pfisterer, and Robert Detrano. Heart Disease. UCI Machine Learning Repository, 1988

    1988

  70. [78]

    Parking Birmingham

    Daniel Stolfi. Parking Birmingham. UCI Machine Learning Repository, 2019

    2019

  71. [79]

    Ames, iowa: Alternative to the boston housing data as an end of semester regression project

    Dean De Cock. Ames, iowa: Alternative to the boston housing data as an end of semester regression project. Journal of Statistics Education , 19, 11 2011. doi:10.1080/10691898.2011.11889627

    2011

  72. [80]

    Combining similarity in time and space for training set formation under concept drift

    Indr˙e Žliobait˙e. Combining similarity in time and space for training set formation under concept drift. Intelligent Data Analysis, 15(4):589–611, 2011

    2011

  73. [81]

    A closer look at in-context learning under distribution shifts

    Kartik Ahuja and David Lopez-Paz. A closer look at in-context learning under distribution shifts. arXiv preprint arXiv:2305.16704, 2023. 16 Table of Contents A Appendix 17 A.1 Reproducibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 A.2 Broader Im...

    2023 arXiv

  74. [84]

    In healthcare, this can ensure diagnostic and prognostic models remain reliable as data shifts over time

    Increased Model Longevity: Our approach extends the usable lifespan of deployed ML models by adapting to distribution shifts, reducing the need for retraining and leading to cost savings and more stable performance. In healthcare, this can ensure diagnostic and prognostic mode...

  75. [85]

    Our Bayesian approach for tackling distribution shift provides a new perspective that can spur further methodological innovations

    Improved Decision Making and Long-Term Predictions: Drift-Resilient TabPFN en- hances decision-making in critical fields like finance and climate science by enabling more robust, longer-term predictions. Our Bayesian approach for tackling distribution shift provides a new pers...

  76. [86]

    Potential Misuse: Like any ML advance, more robust models could be misused for harmful purposes if not developed and deployed responsibly

  77. [87]

    However, we note that these costs are one-time, while the resulting model can be applied with minimal energy usage

    Environmental Cost: While the trained model is now efficiently applicable to various datasets, the considerable computational resources used for initial training should be noted. However, we note that these costs are one-time, while the resulting model can be applied with mini...

  78. [88]

    Infrastructure: The experiments were conducted on an internal SLURM cluster equipped with RTX 2080 TI GPUs and CPUs of type AMD EPYC 7502, 32C/64T, @ 2.50-3.35GHz

  79. [89]

    Baseline Experiments: Each baseline experiment utilized 8 CPUs, 1 GPU, and 62.5 GB RAM, with a hyperparameter optimization (HPO) runtime of 1200 seconds per dataset per split, repeated three times to ensure reliability

  80. [90]

    Pre-training for Drift-Resilient TabPFN and TabPFN-base: These models were pre- trained three times each, requiring 64 CPUs, 8 GPUs, and 500 GB RAM, requiring approxi- mately 7 and 8 days respectively

  81. [91]

    Both TabPFN-base and Drift-Resilient TabPFN underwent preprocessing optimization that utilized 8 CPUs, 1 GPU, and 62.5 GB RAM across 300 runs, each lasting between 0.5 to 1 hour

    Hyperparameter Optimization for Drift-Resilient TabPFN and TabPFN-base: We used 32 CPUs, 4 GPUs, and 250 GB RAM, running approximately 40 configurations and taking about one day per pre-training session for optimizing the hyperparameters of the novel prior-data generating mech...

  82. [92]

    Slowness in traffic (%)

    employs contrastive learning to maximize the agreement between different augmentations of the same image, thereby enhancing the quality of learned visual representations. The approach benefits from learnable nonlinear transformations and optimized contrastive loss parameters. ...

    2009

  83. [93]

    Comprehensive evaluations across 18 synthetic and real-world datasets

    Claims Question: Do the main claims made in the abstract and introduction accurately reflect the paper’s contributions and scope? Answer: [Yes] Justification: We make the following main claims and give justification alongside: • We claim "Comprehensive evaluations across 18 sy...

    2024

  84. [94]

    Conclusions and Limitations

    Limitations Question: Does the paper discuss the limitations of the work performed by the authors? Answer: [Yes] Justification: The paper acknowledges several limitations, including challenges with compu- tational demands when scaling to larger datasets, interpretability, and ...

  85. [95]

    Theory Assumptions and Proofs Question: For each theoretical result, does the paper provide the full set of assumptions and a complete (and correct) proof? Answer: [NA] Justification: The paper does not present new theoretical proofs but focuses on algorithmic development and ...

  86. [96]

    Reproducibility

    Experimental Result Reproducibility Question: Does the paper fully disclose all the information needed to reproduce the main ex- perimental results of the paper to the extent that it affects the main claims and/or conclusions of the paper (regardless of whether the code and da...

  87. [97]

    Reproducibility

    Open access to data and code Question: Does the paper provide open access to the data and code, with sufficient instruc- tions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: We provide the full evaluat...

  88. [98]

    Reproducibility

    Experimental Setting/Details 38 Question: Does the paper specify all the training and test details (e.g., data splits, hyper- parameters, how they were chosen, type of optimizer, etc.) necessary to understand the results? Answer: [Yes] Justification: Comprehensive details rega...

  89. [99]

    We report 95% Confidence Intervals in all our quantitative results and mark this appropriately in the paper

    Experiment Statistical Significance Question: Does the paper report error bars suitably and correctly defined or other appropriate information about the statistical significance of the experiments? Answer: [Yes] Justification: The paper reports error bars and statistical signi...

  90. [100]

    See Section A.3 for details

    Experiments Compute Resources Question: For each experiment, does the paper provide sufficient information on the com- puter resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [Yes] Justification: Details about the compu...

  91. [101]

    Code Of Ethics Question: Does the research conducted in the paper conform, in every respect, with the NeurIPS Code of Ethics https://neurips.cc/public/EthicsGuidelines? Answer: [Yes]

  92. [102]

    Please see Section A.2 for details

    Broader Impacts Question: Does the paper discuss both potential positive societal impacts and negative societal impacts of the work performed? Answer: [Yes] Justification: The paper thoroughly explores the broader impacts of Drift-Resilient TabPFN. Please see Section A.2 for details

  93. [103]

    The research does not involve scraped datasets or applications that directly impact individual privacy or security, thus specific high-risk safeguards are not applicable

    Safeguards Question: Does the paper describe safeguards that have been put in place for responsible release of data or models that have a high risk for misuse (e.g., pretrained language models, image generators, or scraped datasets)? Answer: [NA] Justification: The paper focus...

  94. [104]

    It explicitly mentions the use of datasets, with all necessary citations in Section A.7

    Licenses for existing assets Question: Are the creators or original owners of assets (e.g., code, data, models), used in the paper, properly credited and are the license and terms of use explicitly mentioned and properly respected? Answer: [Yes] Justification: The paper credit...

  95. [105]

    We also provide experimental notebooks for easy usage

    New Assets Question: Are new assets introduced in the paper well documented and is the documentation provided alongside the assets? Answer: [Yes] Justification: We release documentation of the released models and code alongside the paper. We also provide experimental notebooks...

  96. [106]

    Thus, this section is not applicable to the current study

    Crowdsourcing and Research with Human Subjects Question: For crowdsourcing experiments and research with human subjects, does the paper include the full text of instructions given to participants and screenshots, if applicable, as well as details about compensation (if any)? A...

  97. [107]

    Institutional Review Board (IRB) Approvals or Equivalent for Research with Human Subjects Question: Does the paper describe potential risks incurred by study participants, whether such risks were disclosed to the subjects, and whether Institutional Review Board (IRB) approvals...

  98. [1945]

    URL http://www.jstor.org/stable/3001968

    ISSN 00994987. URL http://www.jstor.org/stable/3001968

  99. [2023]

    doi:10.1109/TKDE.2022.3178128

    2022

Pith tools

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