Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

TarDiff: Target-Oriented Diffusion Guidance for Synthetic Electronic Health Record Time Series Generation

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Synthetic EHR time series generated by influence-guided diffusion are claimed to improve downstream clinical classifiers beyond distribution-matching baselines.

desk verdict A genuinely new idea—influence-guided diffusion for EHR time series—with consistent reported gains, but the implemented guidance term doesn't match the derived influence and the empirical reporting needs error bars and a cleaner validation story. read the letter →

arxiv 2504.17613 v1 pith:RJXDRPTE submitted 2025-04-24 cs.LG

Bowen Deng , Chang Xu , Hao Li , Yuhao Huang , Min Hou , Jiang Bian This is my paper · ORCID
classification cs.LG
keywords syntheticEHRtimeseriesdiffusionmodelsinfluencefunctionstarget-orientedgenerationdownstreamtaskutilityclassimbalanceTSTRevaluationclinical
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes TarDiff, a diffusion-based generator for synthetic electronic health record time series whose samples are optimized for their expected effect on a downstream clinical classifier rather than for resemblance to the training data. The central claim is that embedding an influence-function gradient into the reverse diffusion process steers generation toward samples that reduce task-specific loss, and that this yields better mortality and ICU-stay classifiers than samples produced by distribution-matching generators. In Train-on-Synthetic-Test-on-Real experiments, a classifier trained only on TarDiff samples reaches MIMIC-III mortality AUPRC of 0.1799, above the 0.1736 achieved by a classifier trained on real data; across six datasets TarDiff reports gains of up to 20.4% in AUPRC and 18.4% in AUROC over generative baselines. The paper also reports that minority-class samples carry much larger gradient norms and that minority-only guidance more than doubles minority F1 on MIMIC-III (0.163 vs 0.056 real-only), indicating the mechanism can act as a counter to class imbalance. If the claim holds, clinical machine learning gains a way to synthesize training data for rare or underrepresented conditions, not merely a way to imitate the training distribution.

What carries the argument

The central object is the influence-gradient term $J = \nabla_{x_t}(G\cdot\nabla_\phi \ell(x_t,y;\phi^*))$, inserted into the reverse diffusion mean as $\tilde{\mu}_t = \mu_t + w J$. Here $G$ is a single cached vector, the normalized sum of downstream-loss gradients over the guidance set, and $\ell$ is the task loss. The design mirrors classifier guidance but replaces the class log-likelihood gradient with a task-utility signal; because $G$ is independent of $x_t$, the per-step guidance reduces to a dot product plus one gradient computation, which is why sampling remains cheap. The intended effect is to push the denoising trajectory toward regions where a synthetic sample's loss gradient aligns with the guidance-set gradient, i.e., toward samples whose addition to the training set would reduce expected task loss.

What would settle it

Compute the true influence of a batch of TarDiff-generated samples by adding each sample to the training set, retraining the downstream classifier, and measuring the loss change on a fixed held-out guidance set; then rank-correlate those true changes with the paper's predicted $\Delta L_T(\hat{z})$ from Eq. (23). If the correlation is not clearly positive on MIMIC-III mortality, the guidance direction is not optimizing the stated objective.

Watch

Extended reading notes

Core claim

TarDiff's central claim is that the reverse diffusion update can be modified to $\tilde{\mu}_\theta(x_t,y,t) = \mu_\theta(x_t,y,t) + \alpha \nabla_{x_t}\Delta L_T(\hat{z}_t)$, where $\Delta L_T(\hat{z})$ is the expected reduction in downstream task loss caused by adding synthetic sample $\hat{z}=(x,y)$ to the training set. The paper approximates this influence as $\Delta L_T(\hat{z}) = \nabla_\phi \ell(\hat{z};\phi^*)\cdot G$, with $G$ the accumulated (negative) gradient of the downstream loss over a guidance set drawn i.i.d. from the task distribution. In the implemented pipeline, $G$ is cached once from a pretrained downstream model, and at each denoising step the mean update is $\tilde{\mu}_t = \mu_t + w J$, where $J = \nabla_{x_t}(G\cdot\nabla_\phi \ell(x_t,y;\phi^*))$. The paper reports that this produces synthetic time series that, in TSTR mode, beat a real-data-trained classifier on MIMIC-III mortality (AUPRC 0.1799 vs 0.1736) and outperform five generative baselines across six datasets by up to 20.4% AUPRC and 18.4% AUROC, with one-time gradient-caching overhead of 10-167 seconds.

Load-bearing premise

The load-bearing assumption is that the simplified guidance signal actually points toward synthetic samples that improve the downstream model: the exact influence formula in Eq. (21) has a sample-dependent denominator $\|\nabla_\phi \ell(\hat{z};\phi)\|^2$ that Algorithm 1 omits as a constant factor, and the paper provides no experiment verifying that the simplified direction correlates with true leave-one-out retraining effects.

Editorial extensions

If this is right

  • Synthetic-only training with TarDiff can outperform a classifier trained on real data in at least one TSTR setting: MIMIC-III mortality AUPRC 0.1799 vs 0.1736.
  • Augmenting real data with TarDiff samples improves downstream AUROC across most tasks and mix ratios, with the gains generally growing as the synthetic proportion rises from 0.2 to 1.0.
  • Influence guidance naturally shifts generation toward minority-class patterns; TarDiff raises minority F1 from 0.056 to 0.108 on MIMIC-III and from 0.013 to 0.018 on eICU without explicit class weighting.
  • Restricting the guidance set to minority-only samples further raises minority F1 (0.163 on MIMIC-III), while majority-only guidance degrades it (0.066), showing that the guidance direction is controllable.
  • The added cost is modest: one-time downstream training plus gradient caching (10-167 s across datasets) and a per-step overhead ratio $g(L,D)/h(L,D)$ that is small, making the task-oriented guidance practical.

Reading between the lines

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

  • My inference: if the reported gains are real, the same cached-gradient guidance could be ported to non-temporal tabular and image data with a well-defined downstream loss, since nothing in the update rule is specific to time series.
  • My inference: the minority-only guidance results suggest an explicit, tunable selection rule for the guidance set (for instance, choosing the hardest minority examples), which the paper stops short of specifying.
  • My inference: the paper's evidence implies that distributional fidelity alone is an insufficient benchmark for synthetic EHR generation; a utility-first evaluation protocol would report downstream AUPRC/AUROC gains after augmentation, not just distance-to-real-data scores.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 TarDiff, a diffusion-based generative framework for synthetic electronic health record (EHR) time series that aims to generate samples optimized for a downstream clinical task rather than merely for distributional fidelity. The central idea is to use influence functions: the influence of a candidate synthetic sample is defined as the expected reduction in task loss on a guidance set if that sample were added to the training set, and this influence is computed via the gradient inner product between the candidate sample and the cached gradients of the guidance set. This influence gradient is then added to the reverse diffusion mean at each denoising step. The method is evaluated on six datasets (MIMIC-III, eICU, APAVA, ADFD, PTB, TDBrain) in train-on-synthetic/test-on-real (TSTR) and train-on-synthetic-and-real/test-on-real (TSRTR) protocols, reporting AUPRC/AUROC improvements over several GAN, VAE, and diffusion baselines, plus additional analyses on class imbalance, computational overhead, fidelity, and privacy.

Significance. If the central mechanism is valid, the paper makes a valuable contribution: it explicitly targets synthetic data generation toward downstream task utility, which is a recognized weakness of fidelity-only generative models, and it demonstrates consistent gains across multiple clinical datasets and tasks. The paper also provides a useful experimental breadth, including minority-class analysis, guidance-set scale sensitivity, runtime overhead, and privacy metrics. However, the significance is contingent on resolving the mismatch between the influence derivation and the implemented guidance signal, and on providing statistically grounded comparisons; as presented, the evidence does not yet establish that the reported gains arise from the claimed influence objective.

major comments (3)
  1. [Section 3.3 and Algorithm 1] The implemented guidance signal is not the gradient of the influence function defined in Eq. (21). In Eq. (21), the per-sample denominator ||∇φℓ(ˆz;φ)||^2 depends on the candidate sample ˆz, so it cannot be absorbed into a constant vector G as written in Eq. (22). Algorithm 1 precomputes G from Dguide alone (only accumulating per-sample gradients, then normalizing by |D0|) and then uses J = ∇_{x_t}(G·∇φℓ(x_t,y;φ*)) in Eq. (17). Dropping the denominator changes the guidance direction, because the true gradient of Eq. (21) would include a term from differentiating ||∇φℓ(ˆz;φ)||^2 with respect to x_t. Since Eq. (17) is the core contribution, this mismatch must be fixed or explicitly justified (e.g., by showing the simplified direction is a valid approximation of the true influence gradient).
  2. [Section 4.6 and evaluation protocol] The guidance scale w is selected on the Evaluation-Val subset, and then the final performance is reported on the entire validation set, which includes Evaluation-Val. This constitutes a form of validation-set overfitting: the reported figures in Figure 3 and the downstream tables may be optimistically biased because the same data used to choose w are retained in the final evaluation. The paper claims this is 'unbiased,' but the overlap between model selection and evaluation invalidates that claim. The final evaluation should be performed on a held-out test set (or at least on a split disjoint from both Guidance-Val and Evaluation-Val).
  3. [Tables 1, 2, 4, 5 and Figure 2] No error bars, confidence intervals, or seeds are reported for any of the main quantitative results. Several claimed improvements are small (e.g., MIMIC-III mortality AUPRC 0.1799 vs. Real Data 0.1736; PTB AUPRC 0.95435 vs. TimeVAE 0.95092), and without variance estimates it is not possible to assess whether these differences are meaningful. The paper should report mean and standard deviation over at least three independent runs for the primary comparisons.
minor comments (5)
  1. [Section 2.2] The symbol T is used both for the length of the time series and the number of diffusion steps in Section 2.1; this dual use is confusing and should be disambiguated (e.g., L for series length).
  2. [Section 3.3] The symbol ε in Eqs. (18)-(22) is used as a small perturbation magnitude but is never defined; it should be clarified whether it is a fixed scalar, a learning rate, or an infinitesimal, since the magnitude of G in Eq. (22) depends on it.
  3. [Section 4.6 / Figure 3 caption] The text says the right panel reports AUROC on the Evaluation-Val subset, but the Figure 3 caption says 'assessing AUROC performance on the Guidance-Val subset.' These are inconsistent and should be reconciled.
  4. [Appendix A.3, Table 7] Table 7 lists MIMIC-III as having 26,150 samples, while Section 4.1 and Table 6 report 20,920; one of these is a typo and should be corrected.
  5. [Appendix A.4] The text refers to a 'PTBrain dataset,' which appears to be a typo for TDBrain.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor self-referential influence check and validation-split tuning; central test-set claims are externally validated and not circular.

  1. self definitional [Section 3.2 Eq. (17), Section 3.3 Eqs. (21)-(23), Algorithm 1, Section 4.6 Figure 3]
    "Notice that we can denote the gradient accumulation of the guidance set Dguide by G, we can end up with the following equation: ... ΔLT (ẑ) = ∇ϕℓ(ẑ;ϕ)·G ... (b) J←∇ xt [G·∇ϕℓ(xt,yt;ϕ∗)] ... The left panel illustrates sample influence value changes"

    The 'sample influence' plotted in Fig. 3 is the same inner product ∇ℓ(ẑ)·G that Algorithm 1 uses as the guidance objective J, with G cached from Dguide. Increasing the guidance scale w therefore mechanically increases the displayed influence score; the manipulation check is self-referential and cannot validate that the score tracks true leave-one-out loss reduction. In addition, the denominator ||∇ϕℓ(ẑ)||^2 in Eq. (21) depends on the candidate ẑ, so precomputing G without it (Algorithm 1) makes the implemented direction differ from the claimed gradient of ΔLT; this is a correctness flaw rather than a circular prediction, but it reinforces that the 'influence' being optimized is the paper's own definition rather than an externally verified quantity.

full rationale

The paper's headline results (TSTR and TSRTR, Tables 1-5) are evaluated on held-out test sets that are not used to compute the guidance vector G; the guidance set is the validation split, and the downstream TimesNet classifiers are tested on real test data. The reported gains over baselines are therefore empirical and not forced by construction, so the central claim is not circular. However, two mild self-referential elements exist. First, Section 4.6/Figure 3 validates the influence mechanism by plotting the same G·∇ℓ score that is used as the guidance signal, so the observed 'influence increases' are partly by construction rather than an independent check against retraining effects. Second, the paper states that 'throughout all experiments' the standard validation split is the guidance set, and Section 4.6 tunes the guidance scale on a subset of that validation split before reporting final validation performance; this is a mild selection-on-validation issue, though the test-set comparisons are unaffected. There is no load-bearing self-citation chain: the influence-function approximation is attributed to Charpiat et al. and Anand et al., not to the authors' own prior work, and the cited prior papers are not used to forbid alternatives. Overall circularity is low because the principal empirical claims rest on external test-set benchmarks.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central method relies on the validity of a first-order influence approximation and the representativeness of the guidance set; the only fitted hyperparameter is the guidance scale w, which is tuned on the validation split. No new physical entities are introduced.

free parameters (1)
  • Influence guidance scale w = Section 4.6 reports optimal around +1000 for MIMIC/eICU; Appendix D says 100 used in the main experiments
    Controls the strength of the influence gradient in the reverse diffusion update (Eq. 17, Algorithm 1 Step 3c). Selected per task using the Evaluation-Val subset, so it is a free parameter fitted to validation performance.
assumptions (4)
  • domain assumption The parameter-update rule δφ = ε∇φf(x)/||∇φf(x)||^2 from Charpiat et al. (2019) extends correctly to a loss function ℓ when measuring influence on other samples.
    Used in Section 3.3 Eq. (20) without proof; the original result concerns prediction changes, not general loss changes, and the extension assumes small ε and that one gradient step approximates retraining.
  • domain assumption The guidance set D_guide is i.i.d. from the same distribution P as the test data.
    Stated in Section 3.1 and used to define ΔL_T; if the guidance set is not representative, the optimization target is wrong.
  • ad hoc to paper Adding the influence gradient to the reverse diffusion mean (Eq. 17) preserves the conditional data distribution well enough to generate samples from the target distribution.
    The paper borrows classifier-guidance intuition but does not prove that the modified denoising trajectory still samples p_θ(x|y) in a valid way; this is the core design assumption of the method.
  • standard math The first-order Taylor expansion in Eq. (20), ignoring O(||δφ||^2), is accurate for the guidance set.
    Standard Taylor expansion, but the validity depends on δφ being small, which is not guaranteed in practice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TarDiff: Target-Oriented Diffusion Guidance for Synthetic Electronic Health Record Time Series Generation." pith.science (2026). https://pith.science/paper/RJXDRPTE

@misc{pith2026250417613,
  author       = {Pith},
  title        = {Pith review of: TarDiff: Target-Oriented Diffusion Guidance for Synthetic Electronic Health Record Time Series Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RJXDRPTE}},
  note         = {Machine review of arXiv:2504.17613}
}
read the original abstract

Synthetic Electronic Health Record (EHR) time-series generation is crucial for advancing clinical machine learning models, as it helps address data scarcity by providing more training data. However, most existing approaches focus primarily on replicating statistical distributions and temporal dependencies of real-world data. We argue that fidelity to observed data alone does not guarantee better model performance, as common patterns may dominate, limiting the representation of rare but important conditions. This highlights the need for generate synthetic samples to improve performance of specific clinical models to fulfill their target outcomes. To address this, we propose TarDiff, a novel target-oriented diffusion framework that integrates task-specific influence guidance into the synthetic data generation process. Unlike conventional approaches that mimic training data distributions, TarDiff optimizes synthetic samples by quantifying their expected contribution to improving downstream model performance through influence functions. Specifically, we measure the reduction in task-specific loss induced by synthetic samples and embed this influence gradient into the reverse diffusion process, thereby steering the generation towards utility-optimized data. Evaluated on six publicly available EHR datasets, TarDiff achieves state-of-the-art performance, outperforming existing methods by up to 20.4% in AUPRC and 18.4% in AUROC. Our results demonstrate that TarDiff not only preserves temporal fidelity but also enhances downstream model performance, offering a robust solution to data scarcity and class imbalance in healthcare analytics.

Figures

Figures reproduced from arXiv: 2504.17613 by the authors.

Figure 1
Figure 1. Overview of the Influence Guidance Diffusion framework. In Stage 1, we construct task-specific datasets from the original dataset Dtrain and train downstream models fTi In Stage 2, we compute each sample’s gradient-based influence for total influence G based on DTi and fTi . In Stage 3, we leverage influence signals guide the reverse diffusion process with computing ∆LT (ˆz) = ∇ϕℓT (xt, yt; ϕ) · G. All symbols are d… view at source ↗
Figure 2
Figure 2. Comparison of AUROC values for the Mortality and ICU Stay task on the MIMIC III and [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Influence scale analysis conducted by generating samples from the [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Model performance with varying guidance set sizes on PTBrain dataset. [PITH_FULL_IMAGE:figures/full_fig_p018_4.png]
Figure 5
Figure 5. Figure 5: Visualization of negative samples generated by different methods for ICU-Stay on eICU [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]
Figure 6
Figure 6. Figure 6: Visualization of positive samples generated by different methods for ICU-Stay on eICU [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: Visualization of negative samples generated by different methods for Mortality on eICU [PITH_FULL_IMAGE:figures/full_fig_p021_7.png]
Figure 8
Figure 8. Figure 8: Visualization of positive samples generated by different methods for Mortality on eICU [PITH_FULL_IMAGE:figures/full_fig_p021_8.png]
Figure 9
Figure 9. Figure 9: Visualization of negative samples generated by different methods for ICU Stay on MIMIC [PITH_FULL_IMAGE:figures/full_fig_p022_9.png]
Figure 10
Figure 10. Figure 10: Visualization of positive samples generated by different methods for ICU Stay on MIMIC [PITH_FULL_IMAGE:figures/full_fig_p022_10.png]
Figure 11
Figure 11. Figure 11: Visualization of negative samples generated by different methods for Mortality on MIMIC [PITH_FULL_IMAGE:figures/full_fig_p022_11.png]
Figure 12
Figure 12. Figure 12: Visualization of positive samples generated by different methods for Mortality on MIMIC [PITH_FULL_IMAGE:figures/full_fig_p022_12.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Diff-MN: Diffusion Parameterized MoE-NCDE for Continuous Time Series Generation with Irregular Observations

    cs.LG 2026-01 conditional novelty 6.0 of 10

    Diff-MN generates continuous, arbitrary-resolution time series from irregular observations by diffusing MoE-NCDE dynamics weights, reporting consistent wins over KO-VAE and GT-GAN on ten datasets.

Reference graph

Works this paper leans on

24 extracted references · 8 canonical work pages · cited by 1 Pith paper

  1. [1]

    Influence scores at scale for efficient language data sampling

    Nikhil Anand, Joshua Tan, and Maria Minakova. Influence scores at scale for efficient language data sampling. arXiv preprint arXiv:2311.16298,

    arXiv

  2. [8]

    TimeDP: Learning to Generate Multi-Domain Time Series with Domain Prompts

    Yu-Hao Huang, Chang Xu, Yueying Wu, Wu-Jun Li, and Jiang Bian. Timedp: Learning to generate multi-domain time series with domain prompts. arXiv preprint arXiv:2501.05403,

    work page Pith review arXiv

  3. [9]

    Timehr: Image-based time series generation for electronic health records

    Hojjat Karami, Mary-Anne Hartley, David Atienza, and Anisoara Ionescu. Timehr: Image-based time series generation for electronic health records. CoRR, abs/2402.06318,

    arXiv

  4. [10]

    Auto-encoding variational bayes

    Diederik P Kingma. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114,

    arXiv

  5. [13]

    Bridge: Bootstrapping text to control time-series generation via multi- agent iterative optimization and diffusion modelling

    Hao Li, Yu-Hao Huang, Chang Xu, Viktor Schlegel, Ren-He Jiang, Riza Batista-Navarro, Goran Nenadic, and Jiang Bian. Bridge: Bootstrapping text to control time-series generation via multi- agent iterative optimization and diffusion modelling. arXiv preprint arXiv:2503.02445,

    arXiv

  6. [15]

    Llms are not zero-shot reasoners for biomedical information extraction

    Aishik Nagar, Viktor Schlegel, Thanh-Tung Nguyen, Hao Li, Yuping Wu, Kuluhan Binici, and Stefan Winkler. Llms are not zero-shot reasoners for biomedical information extraction. CoRR, abs/2408.12249,

    arXiv

  7. [17]

    Explicit Temporal Embedding in Deep Generative Latent Models for Longitudinal Medical Image Synthesis

    Julian Schön, Raghavendra Selvan, Lotte Nygård, Ivan Richter V ogelius, and Jens Petersen. Explicit temporal embedding in deep generative latent models for longitudinal medical image synthesis. CoRR, abs/2301.05465,

    work page Pith review arXiv

  8. [18]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456,

    arXiv 2011

Show all 24 references
  1. [19]

    Medformer: A multi-granularity patching transformer for medical time-series classification

    Yihe Wang, Nan Huang, Taida Li, Yujun Yan, and Xiang Zhang. Medformer: A multi-granularity patching transformer for medical time-series classification. arXiv preprint arXiv:2405.19363,

    arXiv

  2. [20]

    Time series data augmentation for deep learning: A survey

    Qingsong Wen, Liang Sun, Fan Yang, Xiaomin Song, Jingkun Gao, Xue Wang, and Huan Xu. Time series data augmentation for deep learning: A survey. arXiv preprint arXiv:2002.12478,

    2002 arXiv

  3. [21]

    Timesnet: Temporal 2d-variation modeling for general time series analysis

    Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. Timesnet: Temporal 2d-variation modeling for general time series analysis. arXiv preprint arXiv:2210.02186,

    arXiv

  4. [23]

    For our analysis, we focus on the first 24 hours of hospitalization for each patient, resulting in 20,920 samples

    A Dataset Details A.1 Critical Care EHR Datasets (MIMIC-III and eICU) MIMIC-III[Johnson et al., 2016] is a large, publicly available database comprising de-identified health-related data associated with over 40,000 patients who stayed in critical care units of the Beth Israel ...

    2016

  5. [1977]

    Timevae: A variational auto- encoder for multivariate time series generation

    Abhyuday Desai, Cynthia Freeman, Zuhui Wang, and Ian Beaver. Timevae: A variational auto- encoder for multivariate time series generation. arXiv preprint arXiv:2111.08095,

    arXiv

  6. [2015]

    Data are sampled every 5 minutes, resulting in 288 time steps

    For our analysis, we extract time-series measurements (heart rate, respiratory rate, and oxygen saturation) from the initial 24-hour window of ICU admission. Data are sampled every 5 minutes, resulting in 288 time steps. Each time step includes 3 features, providing a granular...

    2024

  7. [2016]

    Transitioning from real to synthetic data: Quantify- ing the bias in model

    Aman Gupta, Deepak Bhatt, and Anubha Pandey. Transitioning from real to synthetic data: Quantify- ing the bias in model. arXiv preprint arXiv:2105.04144,

    arXiv

  8. [2017]

    Vector quantized time series generation with a bidirectional prior model

    Daesoo Lee, Sara Malacarne, and Erlend Aune. Vector quantized time series generation with a bidirectional prior model. arXiv preprint arXiv:2303.04743,

    arXiv

  9. [2018]

    PULSAR at mediqa-sum 2023: Large language models augmented by synthetic dialogue convert patient dialogues to medical records

    Viktor Schlegel, Hao Li, Yuping Wu, Anand Subramanian, Thanh-Tung Nguyen, Abhinav Ramesh Kashyap, Daniel Beck, Xiao-Jun Zeng, Riza Theresa Batista-Navarro, Stefan Winkler, and Goran Nenadic. PULSAR at mediqa-sum 2023: Large language models augmented by synthetic dialogue conve...

    2023

  10. [2019]

    Diffusion-ts: Interpretable diffusion for general time series generation

    Xinyu Yuan and Yan Qiao. Diffusion-ts: Interpretable diffusion for general time series generation. arXiv preprint arXiv:2403.01742,

    arXiv

  11. [2020]

    Invdiff: Invariant guidance for bias mitigation in diffusion models

    Min Hou, Yueying Wu, Chang Xu, Yu-Hao Huang, Chenxi Bai, Le Wu, and Jiang Bian. Invdiff: Invariant guidance for bias mitigation in diffusion models. CoRR, abs/2412.08480,

    arXiv

  12. [2021]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598,

    arXiv

  13. [2022]

    Malin, Jon Duke, Walter F

    Edward Choi, Siddharth Biswal, Bradley A. Malin, Jon Duke, Walter F. Stewart, and Jimeng Sun. Generating multi-label discrete electronic health records using generative adversarial networks. CoRR, abs/1703.06490,

    arXiv

  14. [2023]

    Team: PULSAR at probsum 2023: PULSAR: pre-training with extracted healthcare terms for summarising patients’ problems and data augmentation with black-box large language models

    Hao Li, Yuping Wu, Viktor Schlegel, Riza Batista-Navarro, Thanh-Tung Nguyen, Abhinav Ramesh Kashyap, Xiao-Jun Zeng, Daniel Beck, Stefan Winkler, and Goran Nenadic. Team: PULSAR at probsum 2023: PULSAR: pre-training with extracted healthcare terms for summarising patients’ prob...

    2023

  15. [2024]

    Controllable financial market generation with diffusion guided meta agent

    Yu-Hao Huang, Chang Xu, Yang Liu, Weiqing Liu, Wu-Jun Li, and Jiang Bian. Controllable financial market generation with diffusion guided meta agent. arXiv preprint arXiv:2408.12991,

  16. [2025]

    Biodiffusion: A versatile diffusion model for biomedical signal synthesis

    Xiaomin Li, Mykhailo Sakevych, Gentry Atkinson, and Vangelis Metsis. Biodiffusion: A versatile diffusion model for biomedical signal synthesis. CoRR, abs/2401.10282, 2024a. Yizhi Li, Ge Zhang, Xingwei Qu, Jiali Li, Zhaoqun Li, Noah Wang, Hao Li, Ruibin Yuan, Yinghao Ma, Kai Zh...

    arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.