Pith. sign in

REVIEW 3 major objections 6 minor 20 references

CausalDiffTab: Mixed-Type Causal-Aware Diffusion for Tabular Data Generation

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read CausalDiffTab claims its diffusion model, regularized by a causal graph learned from the real table, outperforms every baseline on all fidelity, privacy, and downstream metrics — roughly 14% lower error than the previous best.

desk verdict A reasonable incremental heuristic for tabular diffusion, but the C2ST metric is direction-inverted and Eq. 14 doesn't do what the text claims; should go to review with major-revision expectations, not be accepted as is. read the letter →

arxiv 2506.14206 v1 pith:JQP7BMDP submitted 2025-06-17 cs.CL

Jia-Chen Zhang , Zheng Zhou , Yu-Jie Xiong , Chun-Ming Xia , Fei Dai This is my paper · ORCID
classification cs.CL
keywords tabulardatagenerationdiffusionmodelscausaldiscoverydirectedacyclicgraphregularizationmixed-typesyntheticprivacy-preserving
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

CausalDiffTab tries to settle a practical problem: synthetic tabular data that is faithful enough to train on and safe enough to share. The paper claims that a diffusion model for mixed numerical-categorical data improves on every measured axis if it is first told the causal structure of the real table and penalized whenever its predicted noise contradicts that structure. On seven real-world datasets and seven metrics — marginal fidelity, column-correlation fidelity, distinguishability, precision, recall, privacy, and downstream task performance — it reports the best average scores, roughly 14 percent better than the prior state of the art on headline fidelity. A sympathetic reader would care because real tabular data is everywhere but hard to share; if the claim holds, privacy-preserving synthetic copies become closer to drop-in substitutes for the original.

What carries the argument

The central object is the causal matrix $G \in \{0,1\}^{d \times d}$: a binary directed acyclic graph, one entry per ordered pair of feature columns, learned before training by nonlinear NOTEARS (an MLP-based regression with a DAG constraint, Eqs. 4–5) on the one-hot-encoded training table and binarized with threshold $\tau=0.3$ (Eq. 6). It carries the argument because it defines which pairwise predictions the generative model is allowed to align. The matching step is causal pair matching: the denoising network's predicted noise for each column is formed into the outer-product matrix $\hat{\epsilon}_i \hat{\epsilon}_j$ (numerical predictions use the normalized residual $(x_t - \hat{\mu})/\sigma$; categorical predictions use softmax probabilities), the matrix is masked so only entries $(i,j)$ inside $G$ survive, and the mean of the masked entries is minimized as the causal loss (Eq. 14). The second mechanism is the hybrid adaptive weight (Eq. 16), $w_{\text{hybrid}} = \frac{w_{\max}}{2}(e^{-|\Delta L|} + \frac{1}{1+\sigma_{\text{mean}}})$, which couples the penalty's strength to training stability (loss fluctuation $\Delta L$) and to the noise level $\sigma_{\text{mean}}$, implementing the hierarchical prior fusion principle of learning low-level features first and injecting high-level causal semantics later.

What would settle it

Take a synthetic table generated from a known directed acyclic graph with known nonlinear mechanisms and run the full pipeline: check whether NOTEARS recovers the true edges and directions, whether swapping in the true graph changes the reported shape, trend, and privacy numbers, and whether a randomly permuted causal mask still delivers the same gains. If a permuted mask keeps the gains, the causal directions are not the driver and the adaptive weighting schedule alone explains the improvement.

Watch

Extended reading notes

Core claim

The paper's central claim, stated in the abstract and conclusion, is that CausalDiffTab "outperforms baseline methods across all metrics." Concretely, it reports the lowest average shape error (1.17% vs. 1.33% for the previous state of the art, TabDiff, a 13.7% improvement) and trend error (2.10% vs. 2.40%, a 14.3% improvement), the best average $\alpha$-precision (98.24) and $\beta$-recall (50.32), the best average detection score (0.9748), the best privacy score (DCR 53.83%, closer to the ideal 50% than TabDiff's 54.11%), and downstream task results that approach real-data training (Adult AUC 0.914 vs. real 0.927; Diabetes AUC 0.692 vs. real 0.704). The mechanism is a causal matrix extracted from the raw training table by nonlinear NOTEARS and binarized at threshold $\tau=0.3$ (Eq. 6). During denoising, the outer product of the model's per-feature predictions — numerical: the normalized residual $(x_t - \hat{\mu})/\sigma$; categorical: softmax probabilities — is masked by this matrix and minimized as a causal loss (Eq. 14), suppressing predictions that violate the learned directions. The loss is scaled by a hybrid adaptive weight $w_{\text{hybrid}} = \frac{w_{\max}}{2}(e^{-|\Delta L|} + \frac{1}{1+\sigma_{\text{mean}}})$ (Eq. 16) that weakens the causal penalty during high-noise, unstable training and strengthens it as the model stabilizes, an application of the hierarchical prior fusion idea; the ablation shows that a fixed weight degrades shape and trend, and that nonlinear causal extraction outperforms linear extraction.

Load-bearing premise

The load-bearing premise is that the causal graph learned automatically from the training table — nonlinear NOTEARS binarized at threshold $\tau=0.3$ — matches the true causal directions, because the paper's causal checks compare generated data against that same learned graph; a wrong graph would hard-wire spurious correlations as causal and the evaluation would certify them.

Editorial extensions

If this is right

  • Synthetic data from this model can substitute for real data in downstream training: reported MLE scores land close to real-data performance (Adult AUC 0.914 vs. real 0.927; Diabetes AUC 0.692 vs. real 0.704).
  • Privacy protection improves: DCR scores are closer to the ideal 50% than the previous state of the art on four of five datasets, meaning generated records resemble the training set less closely.
  • Causally implausible records drop sharply: in the Adult dataset, Husband-Female combinations fall from 398 (TabDiff) to 244 and Wife-Male from 271 to 147.
  • Fixed-weight causal regularization hurts fidelity, so any future method that injects causal constraints into generative models should follow the adaptive weighting scheme rather than a constant penalty.
  • The ablation's nonlinear-versus-linear comparison supports the paper's premise that tabular causal relations are too complex for a linear regularizer.

Reading between the lines

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

  • A permuted-mask control experiment would isolate what causes the gains: if a randomly shuffled causal matrix preserves the reported improvements, the improvement comes from the adaptive weighting schedule itself, not from causal direction information.
  • The causal checks are measured against the same learned matrix the model was trained to satisfy; auditing the generated data against an independent ground-truth graph (for example, synthetic data from a known DAG) would be a stricter test the paper does not run.
  • The NOTEARS causal-discovery routine is load-bearing but no NOTEARS citation appears in the reference list, and the hierarchical prior fusion principle is invoked without a citation, so the causal component cannot be verified from the paper's bibliography alone.
  • If the learned causal graph is trustworthy, the same masked-diffusion mechanism extends naturally to counterfactual generation and missing-value imputation in mixed-type tables, since the causal matrix supplies a conditional sampling target; the paper cites adjacent work in that direction but implements neither extension.
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 / 6 minor

Summary. The paper proposes CausalDiffTab, a continuous-time diffusion model for mixed numerical/categorical tabular data. A causal graph is learned from the training data with nonlinear NOTEARS, thresholded into a binary mask, and used in a causal-consistency loss that penalizes masked outer products of predicted noise. The regularization weight is adapted during training based on loss fluctuation and noise level. The authors evaluate on seven datasets against GAN, VAE, autoregressive, and diffusion baselines using shape/trend errors, alpha-precision, beta-recall, C2ST, downstream AUC/RMSE, and DCR, and claim state-of-the-art performance across all metrics. The code is publicly released.

Significance. The problem of tabular data generation with causal awareness is relevant, and the adaptive-weighting idea is a reasonable response to training instability. The authors provide a public implementation and extensive benchmarking, which is commendable. However, the current manuscript contains a sign error in the C2ST evaluation and an inconsistency between the causal loss formula and its stated purpose; these issues affect the central claim of universal improvement. With corrections and re-evaluation, the method could be a useful contribution, but the evidence as presented does not establish the headline result.

major comments (3)
  1. [Section 4.4 / Table 6] The C2ST metric is interpreted backwards. C2ST is the accuracy of a classifier trained to distinguish real from synthetic samples; accuracy near 1 means the synthetic data is perfectly detectable and thus indicates worse fidelity, not better. The caption 'Higher scores reflect superior performance' is therefore incorrect. For Adult, CausalDiffTab's 0.9963 is worse than TabDiff's 0.9832, not a 1.31% improvement, and this pattern occurs on most datasets. Because the abstract and conclusion claim wins 'across all metrics', this is a load-bearing error.
  2. [Section 3.3, Eq. (14), Algorithm 1] The causal loss as written minimizes the mean of predicted-noise outer products over the causal mask S. If S is the set of allowed causal edges, this suppresses interactions along those edges, which contradicts the text's claim of 'retaining correlations aligned with the allowed causal directions'. If S is meant to be the complement (non-causal pairs), then Eq. (14), Algorithm 1, and the surrounding text must be corrected to say so. As written, the loss does not implement the described mechanism.
  3. [Section 3.2 and Section 4.4] The causal graph is estimated from the same training data with NOTEARS and then used as both the regularization target and the basis for judging causal plausibility. The paper provides no independent validation of the learned graph (for example, against known edges in the Adult dataset), so the reported causal-consistency gains may be self-confirming. Figure 2 uses the external Husband/Wife constraints, which is a partial check, but this does not validate the full binary thresholded graph or rule out spurious edges.
minor comments (6)
  1. [Section 4.2] The dataset list contains 'Fault' in addition to the seven datasets shown in Table 1; verify the intended dataset names.
  2. [Tables 4 and 5] The captions say 'Higher error rates reflect higher performance', but these are scores, not error rates; the captions should say 'Higher scores'.
  3. [Section 4.4] The text reports a 14.3% average shape improvement across six datasets, but Table 2 shows seven datasets and a 13.7% improvement; the 14.3% figure corresponds to the trend metric in Table 3.
  4. [Equation (13)] The sentence introducing Eq. (13) is incomplete ('increasing the discretization resolution can help approximate a tighter ELBO.'); complete the sentence and integrate it with the equation.
  5. [Section 4.3] The phrase 'We compare MoR with nine popular...' should read 'We compare our method with nine popular...'.
  6. [Algorithm 1 and Eq. (15)] Algorithm 1 returns L_base multiplied by an adaptive weight w, while Eq. (14) and Eq. (15) use a fixed lambda; clarify the relationship between lambda and w_hybrid.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity in the causal-consistency evidence: the learned causal graph defines both the regularization penalty and the success criterion; central benchmark comparisons remain independent.

  1. self definitional [Sec. 3.3, Eq. (14), and Fig. 2]
    "by combining this with a pre-extracted causal matrix, we can measure the inconsistency between the predictions and the known causal structure—specifically, retaining correlations aligned with the allowed causal directions while suppressing those that violate causality. ... Figure 2: ... The results show the number of causally implausible instances (Husband-Female and Wife-Male) in the Adult dataset. Our method significantly reduces the number of such causally implausible cases."

    The mask S in Eq. (14) is built from the thresholded NOTEARS graph G (Eq. 6), which is itself learned from the same Adult training table. The causal loss is explicitly designed to suppress predictions inconsistent with G. Figure 2 then reports that CausalDiffTab reduces Husband-Female/Wife-Male counts, calling them causally implausible. These are the same directed relations encoded in G (e.g., Relationship→Sex), so the reduction is a direct effect of the training objective, not an independent confirmation that the causal structure is correct. The fitted graph simultaneously defines the penalty and the success criterion, so this particular evidence is self-confirming. The shape, trend, MLE, and DCR evaluations do not use the learned graph as their target and remain independent evidence.

full rationale

The main benchmark claim is not circular: shape, trend, α-precision/β-recall, MLE, and DCR are all computed against the real data or downstream outcomes, not against the fitted causal matrix. No load-bearing self-citation chain is present; the paper's self-citations (e.g., Zhang et al., 2025) are background only. The genuine circular element is the causal-consistency demonstration: the NOTEARS graph is estimated from the training data, used as the regularization mask in Eq. (14), and then Figure 2 counts violations of those same learned relations as evidence of causal awareness. That step is self-validating. The C2ST direction issue in Table 6 is a serious correctness/sign problem (higher detection accuracy means more distinguishable, not better) but it is not a circularity and is not scored here; it does, however, weaken the 'all metrics' wording. Overall partial circularity in one supporting evaluation, while the central comparative results retain independent content, warrants a 4.

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

The method depends on a heuristic causal loss (Eq. 14), an adaptive weighting heuristic (Eq. 16), a thresholded causal graph (Eq. 6), and several unreported hyperparameters. The causal graph is estimated from the same training data that is later used to evaluate causal consistency, which is a circularity burden on the causal-awareness claim.

free parameters (5)
  • Binary causal threshold tau = 0.3 (example)
    Used in Eq. 6 to convert the NOTEARS weight matrix into a hard causal mask. No sensitivity analysis is reported.
  • Causal loss weight lambda = not reported
    Global scale of the causal regularization term in Eqs. 14-15. The paper gives no value or schedule.
  • Maximum adaptive weight w_max = not reported
    Scales the hybrid weight in Eq. 16. No default or tuning procedure is given.
  • NOTEARS sparsity alpha and weight decay beta = not reported
    Regularization coefficients in Eq. 4 that determine the learned causal matrix. Values are not reported.
  • EMA smoothing coefficient = not reported
    Used in Algorithm 1 line 12 to estimate loss fluctuation delta L. Not specified.
assumptions (4)
  • domain assumption The NOTEARS-estimated weight matrix A, thresholded at tau, represents the true causal structure of the observed table.
    Observational causal discovery is underdetermined in general. The paper does not validate the estimated edges against external domain knowledge. Section 3.2, Eqs. 4-6.
  • ad hoc to paper Pairwise products of predicted noise values encode causal interaction strength between variables.
    This is the foundation of L_causal in Eq. 14, but no derivation or empirical support is given.
  • ad hoc to paper Minimizing the masked mean of these products enforces the allowed causal directions.
    Algorithm 1 computes Mean(G ⊙ M). The text describes retaining aligned correlations, but the loss as written suppresses all masked correlations.
  • ad hoc to paper The Hierarchical Prior Fusion principle justifies increasing causal regularization as training stabilizes.
    Section 3.4 invokes hierarchical VAEs as an analogy, but Eq. 16 is a heuristic, not a derived objective.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CausalDiffTab: Mixed-Type Causal-Aware Diffusion for Tabular Data Generation." pith.science (2026). https://pith.science/paper/JQP7BMDP

@misc{pith2026250614206,
  author       = {Pith},
  title        = {Pith review of: CausalDiffTab: Mixed-Type Causal-Aware Diffusion for Tabular Data Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JQP7BMDP}},
  note         = {Machine review of arXiv:2506.14206}
}
read the original abstract

Training data has been proven to be one of the most critical components in training generative AI. However, obtaining high-quality data remains challenging, with data privacy issues presenting a significant hurdle. To address the need for high-quality data. Synthesize data has emerged as a mainstream solution, demonstrating impressive performance in areas such as images, audio, and video. Generating mixed-type data, especially high-quality tabular data, still faces significant challenges. These primarily include its inherent heterogeneous data types, complex inter-variable relationships, and intricate column-wise distributions. In this paper, we introduce CausalDiffTab, a diffusion model-based generative model specifically designed to handle mixed tabular data containing both numerical and categorical features, while being more flexible in capturing complex interactions among variables. We further propose a hybrid adaptive causal regularization method based on the principle of Hierarchical Prior Fusion. This approach adaptively controls the weight of causal regularization, enhancing the model's performance without compromising its generative capabilities. Comprehensive experiments conducted on seven datasets demonstrate that CausalDiffTab outperforms baseline methods across all metrics. Our code is publicly available at: https://github.com/Godz-z/CausalDiffTab.

Figures

Figures reproduced from arXiv: 2506.14206 by the authors.

Figure 1
Figure 1. A simple example of causal awareness guidance to data privacy protection and beyond Assefa, Dervovic and et al. (2020); Hernandez, Epelde, Alberdi and et al. (2022). These technologies play a critical role in modern data management and analysis. Traditional statistical models, such as Gaussian mixture models, fit data using simple probability distributions, which limits their expressive capability Borisov, Sessler, … view at source ↗
Figure 2
Figure 2. A comparison of whether causal regularization is included. The results show the number of causally implausible instances (Husband-Female and Wife-Male) in the Adult dataset. Our method significantly reduces the number of such causally implausible cases. and reverse denoising mechanisms. Researchers have been actively exploring ways to extend this powerful framework to tabular data Kim, Lee, Shin and et al. (2022); K… view at source ↗
Figure 3
Figure 3. This section provides a high-level overview of CausalDiffTab. The model constructs a causal correlation matrix by applying one-hot encoding to categorical features, thereby establishing interpretable causal relationships between different feature types . During the reverse denoising process, it dynamically aligns this causal matrix with the noise prediction matrix generated at each diffusion step via a causality-con… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Comparison of training loss results and training evaluation (average of shape and trend) results with and without causal regularization on Shoppers tasks. 0 800 1600 2400 3200 4000 4800 5600 6400 7200 8000 Training Epoch 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 Discriminato…
Figure 5
Figure 5. Figure 5: The results on Beijing tasks. of generated samples and their suitability for downstream tasks, while maintaining data diversity. 4.7.2. Hybrid Adaptive Causal Regularization We further conduct a second ablation study to eval￾uate the effectiveness of the hybrid adaptiv…
Figure 6
Figure 6. Figure 6: Visualization of marginal density comparisons be￾tween generated data and real data. Line plots represent numerical results, while bar charts represent categorical results. Fei Dai: Resources, Software, Supervision, Validation. Declaration of competing interest The aut…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 10 canonical work pages

  1. [3]

    CoRR abs/2311.15127

    Stable video diffu- sion: Scaling latent video diffusion models to large datasets. CoRR abs/2311.15127. URL:https://doi.org/10.48550/arXiv.2311.15127. Borisov, V., Sessler, K., Leemann, T., et al.,

    arXiv doi:10.48550/arxiv.2311.15127

  2. [6]

    doi:10.48550/arXiv.2111.09543

    DeBERTav3: Improving deBERTa us- ingELECTRA-stylepre-trainingwithgradient-disentangledembedding sharing, in: Proceedings of the International Conference on Learning Representations. doi:10.48550/arXiv.2111.09543. Hernandez, M., Epelde, G., Alberdi, A., et al.,

    arXiv doi:10.48550/arxiv.2111.09543

  3. [7]

    Denoising diffusion probabilistic models, in: Advances in Neural Information Processing Systems, Curran Associates, Inc.. pp. 6840–6851. URL: https://proceedings.neurips.cc/paper_files/paper/2020/file/ 4c5bcfec8584af0d967f1ab10179ca4b-Paper.pdf. Karras, T., Aittala, M., Aila, T., et al.,

    work page 2020

  4. [8]

    Karras,T.,Laine,S.,Aittala,M.,etal.,2020

    A style-based generator architecture for generative adversarial networks, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Karras,T.,Laine,S.,Aittala,M.,etal.,2020. Analyzingandimprovingthe imagequalityofstylegan,in:ProceedingsoftheIEEE/CVFConference on Computer Vision and Pattern Recognition (CVPR). Kim,J.,Lee...

    work page 2020

  5. [10]

    URL: https://arxiv.org/abs/1312.6114,arXiv:1312.6114

    Auto-encoding variational bayes. URL: https://arxiv.org/abs/1312.6114,arXiv:1312.6114. Komanduri, A., Zhao, C., Chen, F., Wu, X.,

    arXiv

  6. [13]

    Knowledge-Based Systems 280, 110956

    Generativemtd: A deep synthetic data generation framework for small datasets. Knowledge-Based Systems 280, 110956. URL:https: //www.sciencedirect.com/science/article/pii/S0950705123007062, doi:https://doi.org/10.1016/j.knosys.2023.110956. Song,J.,Meng,C.,Ermon,S.,2021a. Denoisingdiffusionimplicitmodels, in:InternationalConferenceonLearningRepresentations....

    work page arXiv

  7. [14]

    URL: https://proceedings.neurips.cc/paper_files/paper/2019/file/ 3001ef257407d5a371a96dcd947c7d93-Paper.pdf

    Generative modeling by estimating gradients of the data distribution, in: Advances in Neural Information Processing Systems, Curran Associates, Inc. URL: https://proceedings.neurips.cc/paper_files/paper/2019/file/ 3001ef257407d5a371a96dcd947c7d93-Paper.pdf. Song,Y.,Sohl-Dickstein,J.,Kingma,D.P.,etal.,2021b.Score-basedgener- ativemodelingthroughstochasticd...

    work page 2019

  8. [15]

    URL:https://arxiv.org/abs/2307.09288, arXiv:2307.09288

    Llama 2: Open foundation and fine-tuned chat models. URL:https://arxiv.org/abs/2307.09288, arXiv:2307.09288. Uemura, K., Takagi, T., Takayuki, K., et al.,

    arXiv

Show all 20 references
  1. [16]

    URL: https://proceedings.neurips.cc/paper_files/paper/2019/file/ 254ed7d2de3b23ab10936522dd547b78-Paper.pdf

    Modeling tabular data using conditional gan, in: Advances in Neural Information Processing Systems, Curran Associates, Inc. URL: https://proceedings.neurips.cc/paper_files/paper/2019/file/ 254ed7d2de3b23ab10936522dd547b78-Paper.pdf. Yan, J., Huang, H., Yang, K., et al.,

    2019

  2. [18]

    URL:https://arxiv

    Editworld: Simulating world dynamicsforinstruction-followingimageediting. URL:https://arxiv. org/abs/2405.14785,arXiv:2405.14785. You,J.,Ma,X.,Ding,Y.,Kochenderfer,M.J.,Leskovec,J.,2020. Handling missing data with graph representation learning. Advances in Neural Information P...

    2020 arXiv

  3. [19]

    Parameter-efficient fine- tuning of large language models via deconvolution in subspace, in: Proceedingsofthe31stInternationalConferenceonComputationalLin- guistics, Association for Computational Linguistics, Abu Dhabi, UAE. pp. 3924–3935. URL:https://aclanthology.org/2025.col...

    2025

  4. [20]

    URL:https://openreview.net/forum?id=4q9kFrXC2Ae

    Diffusion models for missing value imputation in tabular data, in: NeurIPS 2022 First Table Representation Workshop. URL:https://openreview.net/forum?id=4q9kFrXC2Ae. CV Radhakrishnan et al.:Preprint submitted to ElsevierPage 12 of 12

    2022

  5. [32]

    Rombach,R.,Blattmann,A.,Lorenz,D.,Esser,P.,Ommer,B.,2022

    URL: https://proceedings.neurips.cc/paper_files/paper/2019/file/ bdbca288fee7f92f2bfa9f7012727740-Paper.pdf. Rombach,R.,Blattmann,A.,Lorenz,D.,Esser,P.,Ommer,B.,2022. High- resolutionimagesynthesiswithlatentdiffusionmodels,in:Proceedings of the IEEE/CVF Conference on Computer ...

    2019

  6. [2017]

    URL: https://arxiv.org/abs/1701.07875,arXiv:1701.07875

    Wasserstein gan. URL: https://arxiv.org/abs/1701.07875,arXiv:1701.07875. Assefa, S.A., Dervovic, D., et al.,

    arXiv

  7. [2019]

    BERT: Pre-training of deep bidirectional transformers for language understanding, in: Proceedings of the 2019 Conference of the North American Chapter of the Asso- ciation for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), Association...

    2019 doi

  8. [2020]

    URL:https://doi.org/10.1145/ 3422622, doi:10.1145/3422622

    Generative adversarial networks 63, 139–144. URL:https://doi.org/10.1145/ 3422622, doi:10.1145/3422622. He, P., Gao, J., Chen, W.,

    doi

  9. [2022]

    Sos: Score-based oversampling for tabular data, in: Proceedings of the 28th ACM SIGKDD Conference on KnowledgeDiscoveryandDataMining,AssociationforComputingMa- chinery. p. 762–772. URL:https://doi.org/10.1145/3534678.3539454, doi:10.1145/3534678.3539454. CV Radhakrishnan et al...

  10. [2023]

    arXiv:2303.08774

    Gpt-4 technical report. arXiv:2303.08774. Arjovsky, M., Chintala, S., Bottou, L.,

    arXiv

  11. [2024]

    URL:https://www.sciencedirect.com/science/article/pii/ S0141938223002615, doi:https://doi.org/10.1016/j.displa.2023.102627

    Codp-1200: An aigc basedbenchmarkforassistinginchildlanguageacquisition.Displays82, 102627. URL:https://www.sciencedirect.com/science/article/pii/ S0141938223002615, doi:https://doi.org/10.1016/j.displa.2023.102627. Li,J.,Qian,R.,Tan,Y.,etal.,2024. Tabsal:Synthesizingtabularda...

    2023

  12. [2025]

    Knowledge-Based Systems 309, 112899

    Synthetic data for enhanced privacy: A vae-gan approach against membership inference attacks. Knowledge-Based Systems 309, 112899. URL:https: //www.sciencedirect.com/science/article/pii/S0950705124015338, doi:https://doi.org/10.1016/j.knosys.2024.112899. Yang, L., Zeng, B., Li...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.