Pith. sign in

REVIEW 4 major objections 6 minor 35 references

A generalized Wasserstein-2 distance approach for efficient reconstruction of random field models using stochastic neural networks

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper proves that stochastic neural networks can approximate any sufficiently regular random field with mixed continuous and categorical outputs to arbitrary accuracy under a generalized Wasserstein-2 distance, and trains them with a…

desk verdict A plausible extension of local W2 training to mixed continuous/categorical random variables, but the central approximation theorem depends on an unstated prior theorem and hand-waved regularity conditions; worth a serious referee who expects revision. read the letter →

arxiv 2507.05143 v1 pith:KK5MY6CB submitted 2025-07-07 cs.LG

Mingtao Xia , Qijing Shen This is my paper · ORCID
classification cs.LG MSC 60A0568Q8765C99
keywords WassersteindistanceuncertaintyquantificationrandomfieldreconstructionmixedvariablestochasticneuralnetworkuniversalapproximationlocalsquaredWasserstein-2loss
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

This paper proposes a way to train stochastic neural networks so that they reproduce the full conditional distribution of a random field model whose output mixes continuous and categorical components, not just its mean. The central theoretical claim is a universal approximation result: under boundedness, Lipschitz, and regularity conditions, an SNN can make the integrated squared generalized Wasserstein-2 distance from its output distribution to the true field smaller than any prescribed positive tolerance. To make this usable from finite data, the paper defines a generalized Wasserstein-2 distance whose categorical part is a bounded discrepancy, introduces a differentiable surrogate so gradients flow through discrete targets, and proves a finite-sample bound showing that a local empirical version of the loss converges to the true distance. In experiments on classification, a mixed abalone dataset, and a stochastic gene-toggle dynamical system, the trained SNN matches or improves on comparison methods for uncertainty quantification.

What carries the argument

The load-bearing object is a modified norm on the mixed vector, $\|y\|^2=\lambda\sum_{i=1}^{d_1} y_i^2 + \sum_{j=d_1+1}^d \hat\delta_{y_j,0}$, where $\hat\delta_{y_j,0}=4y_j^2$ for $|y_j|\le 1/2$ and $1$ otherwise. This makes categorical mismatches contribute a bounded, almost quadratic penalty instead of an unbounded Euclidean distance, and it defines the generalized Wasserstein-2 distance $\hat W_2$ between conditional distributions. The approximation proof smooths the categorical components with a mollifier so that the known SNN universal approximation theorem for continuous random fields applies, then transfers the resulting bound back to the original mixed measure by triangle inequalities. The training side uses a local empirical version of $\hat W_2^2$, with a differentiable surrogate for the categorical discrepancy, quadratic when prediction is near the label and cosine-based when it is far, so that gradient descent can train through discrete targets.

What would settle it

Take a mixed random field with bounded outcomes whose categorical class probabilities jump discontinuously at a threshold in $x$, so the uniform Lipschitz condition in Eq. (2.8) fails, and train an SNN with the proposed local squared loss; if the learned distributions still converge to the target under $\hat W_2$, the Lipschitz assumption is unnecessary, and if they do not, the theorem's stated scope is confirmed. A complementary check is to verify the technical regularity conditions on $D$ named in [34, Appendix H] for a concrete domain used in the experiments; if those conditions are violated, the proof of Theorem 2.1 is incomplete as written.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is Theorem 2.1: for any random field $y_x=y(x;\omega)$ whose conditional distributions $f_x$ are uniformly bounded, uniformly Lipschitz in $x$ under the generalized Wasserstein-2 metric, and satisfy the smoothness conditions in Assumption 2.1, and for any $\epsilon_1>0$, there exists a stochastic neural network whose output distribution $\hat f_x$ satisfies $\int_D \hat W_2^2(f_x,\hat f_x)\,\nu(dx)\le \epsilon_1$. Continuous outputs are rounded at test time to deliver categorical predictions, while training uses a surrogate of the categorical discrepancy that keeps gradients nonzero. The paper also proves Theorem 2.2, a finite-sample generalization bound showing that the local empirical loss approaches the true squared distance at a rate controlled by $1/\sqrt N$, the neighborhood size $\delta$, and a dimension-dependent empirical-measure term. These two results underwrite the practical recipe: minimize the generalized local squared Wasserstein-2 loss over the means and variances of the stochastic network weights.

Load-bearing premise

The load-bearing premise is the regularity package in Assumption 2.1, above all the uniform Lipschitz continuity of the true conditional distributions in $x$ and the technical conditions on the domain $D$ that the proof inherits from an earlier continuous-field approximation theorem and simply assumes hold, so that the black-box universal approximation result can be applied.

Editorial extensions

If this is right

  • If Theorem 2.1 is correct, SNNs are a universal model class for mixed-output random fields: no separate continuous and categorical architectures are needed for distributional reconstruction.
  • The local squared loss provides a finite-data objective whose error relative to the true distance is controlled by three explicit terms, sample size, local neighborhood statistics, and Lipschitz regularity, so users can predict when reconstruction will be accurate.
  • Because the loss is differentiable and local, it can be used as a drop-in replacement for mean squared error or cross-entropy in uncertainty-quantification pipelines for mixed targets.
  • The learned per-input randomness comes from sampling stochastic weights, so repeated forward passes at a fixed $x$ give a Monte Carlo estimate of the full conditional distribution.
  • For multivariate categorical targets, encoding them as a single categorical variable improves both accuracy and runtime, matching the dimension dependence in the bound.

Reading between the lines

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

  • The categorical penalty saturates at 1 for well-separated labels, so the loss effectively ignores how different two wrong categories are; permuting the category codes should leave training unchanged, a testable consequence the paper does not run.
  • The mollification step is a general transfer principle: any universal approximation theorem for continuous random fields should lift to mixed random fields under the same assumptions, so the proof strategy could be reused for other stochastic network architectures without redoing the argument.
  • The bound in Theorem 2.2 has three competing terms depending on the neighborhood size $\delta$; balancing $1/\sqrt N$, $h(N(x,\delta),d)$, and $L\delta$ would give a principled rule for choosing $\delta$, which the paper leaves to heuristic search.
  • The method only needs local empirical distributions of outputs, so it should extend to online or streaming settings where minibatches are refreshed during training, though the paper does not analyze this regime.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes a generalized Wasserstein-2 distance for random field models whose outputs contain both continuous and categorical components. It defines a generalized squared W2 loss, proves a universal approximation theorem for stochastic neural networks under this distance (Theorem 2.1) and a generalization error bound for the empirical local loss (Theorem 2.2), then introduces a differentiable surrogate loss and tests the method on classification, mixed continuous/categorical regression (abalone), and a gene-regulatory ODE/Markov jump system.

Significance. If the theoretical results are made fully rigorous, the paper offers a useful principled extension of local squared Wasserstein training to categorical and mixed outputs, a setting where the usual W2 distance is not directly applicable. The differentiable surrogate with a detach-based rounding trick is a practical contribution, and the three diverse UQ experiments plus the sensitivity analyses give useful empirical evidence. The main limitation is that the central approximation theorem currently rests on an unstated external theorem and a smoothing construction that is not fully justified, so the theoretical support does not yet match the strength of the claims.

major comments (4)
  1. [Appendix A, Eq. (A.9)] The proof of Theorem 2.1 reduces the approximation of the smoothed measure f_{ε,x} to the universal approximation theorem of [34, Appendix H], but that theorem is neither stated nor proved in this manuscript, and the text explicitly says 'For simplicity, we assume those conditions hold here' concerning technical regularity conditions on D. Since [34] is a preprint and its hypotheses are not checkable from the present paper, Theorem 2.1 is conditional at its load-bearing step. Please state the external theorem in full and either prove it or verify all of its hypotheses, or replace the reduction with a self-contained proof.
  2. [Appendix A, Eqs. (A.2)-(A.4)] The definition of f_{ε,x} in Eq. (A.2) is not a probability density as written: it appears to depend on a single discrete y with no summation over the categorical support, and it is not normalized. Moreover, the convolution with φ_ε acts only on the categorical coordinates, so the claimed smoothness of f_{ε,x} in all d coordinates does not follow from the uniform continuity of f_x in the continuous components stated in Assumption 2.1.7. This affects the verification of Eqs. (A.4) and (A.9). Please define f_{ε,x} as a genuine convolution in all coordinates (for example, summing over y_2 and mollifying the continuous coordinates as well) and prove that the mixed-derivative bounds in Assumption 2.1 are inherited.
  3. [Section 2, Eqs. (2.1)-(2.2); Appendix B, Eq. (B.7)] The proof of Theorem 2.2 applies the triangle inequality to \hat W_2, but the cost defined in Eq. (2.1) is not shown to make \hat W_2 a metric and it is not a norm: for a one-dimensional continuous component, \|(0.8,0,\dots,0)\| = 1, while 2\|(0.4,0,\dots,0)\| = 1.6, so homogeneity fails. The paper calls Eq. (2.1) a norm and claims the coefficient 4 ensures this, but that claim is false. A lemma establishing that the generalized distance satisfies the triangle inequality (or a different argument) is needed before Eq. (B.7) can be used.
  4. [Section 2.3, Eqs. (2.23)-(2.26)] The loss actually minimized in all experiments is the differentiable surrogate |\cdot|_1 with the detach/straight-through estimator in Eq. (2.25), while Theorems 2.1 and 2.2 are stated for the generalized \hat W_2 built from Eq. (2.1). The equality between |y-\hat y|_1 and \|y-\hat y\| holds only when both variables are rounded to integers, which is not the training phase. No argument shows that minimizing the surrogate controls \hat W_2 or that the detach operation preserves the objective. Please either prove a relation between the surrogate and \hat W_2, or explicitly present Eq. (3.1) as a heuristic whose validation is empirical.
minor comments (6)
  1. [Section 2, Eq. (2.2)] The notation \hat\delta_{y_j,0} has one argument, but Eqs. (2.14)-(2.15) and (D.1) use a two-argument form \hat\delta_{y_i,\hat y_i}; please define this two-argument version explicitly.
  2. [Section 2, Eq. (2.1)] The notation \|\cdot\|_2 is used both for the generalized quantity in Eq. (2.1) and for the Euclidean ℓ2 norm in Eq. (2.7), which is confusing; please use distinct symbols.
  3. [Section 3, Examples 3.1-3.4] Assumptions 2.1.3 and 2.1.4 require the conditional distributions to be uniformly Lipschitz in x under the generalized W2 distance, but the experiments provide no diagnostic or discussion of whether this regularity condition is plausible for the tested models.
  4. [Section 3, Figures 2-4 and Table 2] All experimental results appear to be single-run and no error bars, confidence intervals, or multiple-seed variability are reported, which makes it difficult to judge whether the observed differences between methods are significant.
  5. [Section 3.2, paragraph after Eq. (3.3)] The sentence 'For predicting the categorical sex variable on the testing set' appears to be a copy-paste from Example 3.3; in Example 3.2 the target is a multilabel binary vector, not a sex variable.
  6. [Section 2.3, Eq. (2.25)] The detach operation in round_1 is a straight-through estimator and deserves a more explicit explanation, since it is central to the practical training procedure but is mentioned only in one sentence.

Circularity Check

1 steps flagged · score 8.0 of 10

Theorem 2.1's approximation guarantee is inherited from the authors' own prior preprint [34, Appendix H], whose hypotheses are not verified and whose conditions on D are assumed; the central claim therefore reduces to a self-citation chain.

  1. self citation load bearing [Appendix A, proof of Theorem 2.1, after Eq. (A.9)]
    "Combining Eqs. (A.4), (A.5), the uniform continuity of fε,x(z), and the Lipschitz condition Eq. (A.8), the assumptions in the universal approximation ability theorem of SNNs in [34, Appendix H] hold. Therefore, for any ε0 > 0, from [34, Appendix H], there exists an SNN such that: (A.9) ... (note: [34, Appendix H] also imposes some technical regularity conditions on the bounded set D for x in Eq. (1.1). For simplicity, we assume those conditions hold here.)"

    Theorem 2.1 is the paper's central first-principles claim. Its proof does not establish the existence of the approximating SNN; instead it imports that existence from [34, Appendix H], a preprint by the same authors (Xia, Shen, Maini, Gaffney, Mogilner) that is not stated, proved, or machine-checked here. The new mixed-variable content only smooths the categorical coordinates in (A.2) and verifies inherited regularity properties; the decisive step (A.9) is exactly the continuous-variable universal approximation theorem of the authors' own prior work. The appended note then assumes away unstated technical conditions on D. Thus the central approximation theorem reduces to a self-citation chain: if [34, Appendix H] fails or its hypotheses are not met, Theorem 2.1 has no proof in this paper.

full rationale

I examined the claimed derivation chain for self-definitional reductions, fitted inputs presented as predictions, and renamed known results. No empirical benchmark reduces to a fitted constant: the loss function in Eq. (3.1) is an empirical surrogate for the proposed generalized W2 distance, and the tests use independent permutation tests, R2 statistics, and classification accuracy rather than the training loss itself. The generalization bound in Theorem 2.2 is derived from external quantitative bounds (Fournier-Guillin, the triangle inequality for Wasserstein distances) and does not assume its own conclusion. The one load-bearing circularity is Theorem 2.1. The proof's only mechanism for producing an SNN is the citation to [34, Appendix H], the authors' own prior preprint, which is never reproduced. Moreover, the proof explicitly assumes unstated 'technical regularity conditions' on the bounded set D, converting the stated theorem into a conditional statement whose hypotheses are not fully checked. Because the central theoretical result is forced through this self-citation chain, the circularity score is 8 rather than a lower score. The independent numerical content in Section 3 is not circular, but it does not rescue the theorem's derivation from its reliance on the unverified external self-citation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on data-dependent hyperparameters λ and δ, on the ad hoc coefficient 4 in the truncated quadratic cost, and on two external theorems: the authors' prior SNN universal approximation theorem and standard empirical-process bounds. No new physical or model entity is postulated.

free parameters (3)
  • λ (weight for continuous components in Eq. 2.1) = Σ Var[y_i] for Example 3.3; time-averaged variance for Example 3.4; not used in 3.1-3.2
    Hand-chosen data-dependent scale balancing continuous and categorical terms; central to the definition of the generalized W2 distance.
  • δ (neighborhood radius in loss Eq. 2.20/3.1) = 0.025, 0.5√8, 0.3√7, 0.02 per example
    Controls the local empirical measure; chosen by sensitivity and affects reconstruction accuracy.
  • coefficient 4 in Ŵδ (Eq. 2.2) = 4
    Chosen by hand and by sensitivity tests; the paper claims other constants break the norm property, but even 4 does not yield a true norm.
assumptions (4)
  • domain assumption Assumption 2.1: uniform boundedness of y_x and Ŵ_y_x; independence of ω and W̆ω from x; uniform Lipschitz continuity of f_x and Ŵ_f_x under generalized W2; finiteness of mixed Sobolev norms; uniform continuity of f_x in continuous components.
    Invoked throughout Section 2 and Appendix A to make the universal approximation and generalization proofs go through; not verified on any experiment.
  • domain assumption Universal approximation theorem for SNNs on continuous random fields ([34, Appendix H])
    Used directly in Appendix A, Eq. (A.9), to obtain an SNN approximating the smoothed continuous distribution f_{ε,x}; the present paper assumes its technical conditions on D without checking them.
  • standard math Fournier-Guillin convergence rate for empirical measures in Wasserstein distance ([6, Theorem 1])
    Used in Appendix B, Eq. (B.11)-(B.12), to bound the empirical estimation error of the local measures.
  • standard math Triangle inequality for (generalized) Wasserstein distances ([4, Proposition 2.1])
    Used in Appendix B, Eq. (B.7), to decompose the generalization error.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A generalized Wasserstein-2 distance approach for efficient reconstruction of random field models using stochastic neural networks." pith.science (2026). https://pith.science/paper/KK5MY6CB

@misc{pith2026250705143,
  author       = {Pith},
  title        = {Pith review of: A generalized Wasserstein-2 distance approach for efficient reconstruction of random field models using stochastic neural networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KK5MY6CB}},
  note         = {Machine review of arXiv:2507.05143}
}
read the original abstract

In this work, we propose a novel generalized Wasserstein-2 distance approach for efficiently training stochastic neural networks to reconstruct random field models, where the target random variable comprises both continuous and categorical components. We prove that a stochastic neural network can approximate random field models under a Wasserstein-2 distance metric under nonrestrictive conditions. Furthermore, this stochastic neural network can be efficiently trained by minimizing our proposed generalized local squared Wasserstein-2 loss function. We showcase the effectiveness of our proposed approach in various uncertainty quantification tasks, including classification, reconstructing the distribution of mixed random variables, and learning complex noisy dynamical systems from spatiotemporal data.

Figures

Figures reproduced from arXiv: 2507.05143 by the authors.

Figure 1
Figure 1. An example of the structure of the neural network model used in this study. In the neural network model, for each input x, the weights wi,j,k ∼ N (ai,j,k, σ2 i,j,k) are independently sampled. ReLU means the ReLU activation function and may be replaced with other activation functions. Two structures of forward propagation may be used: the normal feedforward structure or the Resnet [12] structure. Note that the output… view at source ↗
Figure 2
Figure 2. (a) ground truth yx versus yˆx generated by the trained SNN (for visualization clarity, we scatter (x, yx) and (x, yˆx + 0.1)). The number of training data is N = 1000. (b) the permutation test null hypothesis rejection rate w.r.t. the number of training data points. In (a)(b), σ = 0.4 in Eq. (3.2). (c) the permutation rejection rate w.r.t. the uncertainty level σ in Eq. (3.2) (the number of training data N = 1000).… view at source ↗
Figure 3
Figure 3. (a) the ground truth and predicted values of the continuous variable rings (y1 and yˆ1) in the neighborhoods of ten randomly chosen samples which have no fewer than 50 neighbors in the neighborhood kx − x˜k2 ≤ δ in the testing set. (b) the mean and standard deviations of the ground truth and predicted values of the continuous variable rings (y1 and yˆ1) in the neighborhoods of ten randomly chosen samples which have … view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: (a, b) the scaled mRNA counts transcripted from gene 1 and gene 2 over time, respectively. (c, d) mean and standard deviation of the rate of change for the scaled mRNA counts associated with gene 1 and gene 2 over time. (e, f ) ground truth and predicted ratios of cell…
Figure 5
Figure 5. Figure 5: (a) the p-value test rejection rate w.r.t. the number of hidden layers in the SNN. The forward propagation mode is either the Normal mode or the ResNet model, as described in [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 30 canonical work pages

  1. [34]

    M. Xia, Q. Shen, P. Maini, E. Gaffney, and A. Mogilner , A new local time-decoupled squared wasserstein-2 method for training stochastic neural netwo rks to reconstruct uncertain parameters in dynamical systems, arXiv preprint arXiv:2503.05068, (2025)

    arXiv 2025

  2. [1]

    Agresti , An Introduction to Categorical Data Analysis , Wiley, 3rd ed., 2018

    A. Agresti , An Introduction to Categorical Data Analysis , Wiley, 3rd ed., 2018

    work page 2018

  3. [2]

    G. Arya, M. Schauer, F. Sch ¨afer, and C. Rackauckas , Automatic differentiation of programs with discrete randomness, Adv. Neural Inf. Process., 35 (2022), pp. 10435–10447

    work page 2022

  4. [3]

    D. J. Bartholomew, M. Knott, and I. Moustaki , Latent Variable Models and Factor Analysis , Wiley, 7th ed., 2011

    work page 2011

  5. [4]

    Clement and W

    P. Clement and W. Desch , An elementary proof of the triangle inequality for the Wasse rstein metric , Proc. Amer. Math. Soc., 136 (2008), pp. 333–339

    work page 2008

  6. [5]

    Cuturi , Sinkhorn distances: lightspeed computation of optimal tra nsport, Adv

    M. Cuturi , Sinkhorn distances: lightspeed computation of optimal tra nsport, Adv. Neural Inf. Process., 26 (2013)

    work page 2013

  7. [6]

    Fournier and A

    N. Fournier and A. Guillin , On the rate of convergence in Wasserstein distance of the emp irical measure, Probab. Theory Relat. Fields, 162 (2015), pp. 707–738

    work page 2015

  8. [7]

    L. M. Friedman, C. D. Furberg, D. L. DeMets, D. M. Reboussin, a nd C. B. Granger , Funda- mentals of clinical trials , Springer, 2015

    work page 2015

Show all 35 references
  1. [8]

    Gao and M

    Y. Gao and M. K. Ng , Wasserstein generative adversarial uncertainty quantific ation in physics-informed neural networks, J. Comput. Phys., 463 (2022), p. 111270

    2022

  2. [9]

    Ghahramani , Bayesian nonparametrics and the probabilistic approach to modelling, Phil

    Z. Ghahramani , Bayesian nonparametrics and the probabilistic approach to modelling, Phil. Trans. R. Soc. A, 371 (2013), p. 20110553. This manuscript is for review purposes only. A GENERALIZED WASSERSTEIN-2 DISTANCE APPROACH 23

    2013

  3. [10]

    Ghahramani and M

    Z. Ghahramani and M. Jordan , Supervised learning from incomplete data via an EM approach , Adv. Neural Inf. Process., 6 (1993)

    1993

  4. [11]

    Hastie, R

    T. Hastie, R. Tibshirani, and J. Friedman , The Elements of Statistical Learning , Springer, 2nd ed., 2009

    2009

  5. [12]

    K. He, X. Zhang, S. Ren, and J. Sun , Deep residual learning for image recognition , in CVPR, 2016, pp. 770–778

    2016

  6. [13]

    Jaruszewicz and T

    J. Jaruszewicz and T. Lipniacki , Toggle switch: noise determines the winning gene , Phys. Biol., 10 (2013), p. 035007

    2013

  7. [14]

    Q. Jin, X. Luo, Y. Shi, and K. Kita , Image generation method based on improved condition GAN , in ICSAI, IEEE, 2019, pp. 1290–1294

    2019

  8. [15]

    Koller and N

    D. Koller and N. Friedman , Probabilistic Graphical Models: Principles and Technique s, MIT Press, 2009

    2009

  9. [16]

    Lakshminarayanan, A

    B. Lakshminarayanan, A. Pritzel, and C. Blundell , Simple and scalable predictive uncertainty estimation using deep ensembles , in Adv. Neural Inf. Process., vol. 30, 2017

    2017

  10. [17]

    C. E. McCulloch, S. R. Searle, and J. M. Neuhaus , Generalized, Linear, and Mixed Models , Wiley, 2nd ed., 2008

    2008

  11. [18]

    MCF ADDEN, Conditional logit analysis of qualitative choice behavior , Front

    D. MCF ADDEN, Conditional logit analysis of qualitative choice behavior , Front. Econom., (1974)

    1974

  12. [19]

    Mullachery, A

    V. Mullachery, A. Khera, and A. Husain , Bayesian neural networks , arXiv preprint arXiv:1801.07710, (2018)

    2018 arXiv

  13. [20]

    W. Nash, T. Sellers, S. Talbot, A. Cawthorn, and W. Ford , Abalone. UCI Machine Learning Repository, 1994. DOI: https://doi.org/10.24432/C55C7W

    1994 doi

  14. [21]

    A. S. Now ak and K. R. Collins , Reliability of Structures , CRC press, 2012

    2012

  15. [22]

    NYU, NYU HPC hardware specs , 2025

    2025

  16. [23]

    Pesarin and L

    F. Pesarin and L. Salmaso , Permutation Tests for Complex Data: Theory, Applications a nd Software, John Wiley & Sons, 2010

    2010

  17. [24]

    Schiebinger, J

    G. Schiebinger, J. Shu, M. Tabaka, et al. , Optimal-transport analysis of single-cell gene expressio n identifies developmental trajectories in reprogramming , Cell, 176 (2019), pp. 928–943.e22

    2019

  18. [25]

    Sensoy, L

    M. Sensoy, L. Kaplan, and M. Kandemir , Evidential deep learning to quantify classification uncer- tainty, Adv. Neural Inf. Process., 31 (2018)

    2018

  19. [26]

    B. W. Sil verman, Density Estimation for Statistics and Data Analysis , Chapman & Hall, 1986

    1986

  20. [27]

    Solomon, F

    J. Solomon, F. De Goes, G. Peyr ´e, M. Cuturi, A. Butscher, A. Nguyen, T. Du, and L. Guibas , Convolutional Wasserstein distances: Efficient optimal tra nsportation on geometric domains , ACM Trans. Graph., 34 (2015), pp. 1–11

    2015

  21. [28]

    Tian and K

    T. Tian and K. Burrage , Stochastic models for regulatory networks of the genetic to ggle switch , Proc. Natl. Acad. Sci., 103 (2006), pp. 8372–8377

    2006

  22. [29]

    Villani et al

    C. Villani et al. , Optimal Transport: Old and New , vol. 338, Springer, Heidelberg, 2009

    2009

  23. [30]

    W ang, J

    J. W ang, J. Wu, X. Huang, and Z. Xiong , Improved WGAN for image generation methods , in ICMNM, Springer, 2023, pp. 199–211

    2023

  24. [31]

    M. Xia, X. Li, Q. Shen, and T. Chou , An efficient Wasserstein-distance approach for reconstruct ing jump-diffusion processes using parameterized neural networ ks, Mach. Learn.: Sci. Technol., 5 (2024), p. 045052

    2024

  25. [32]

    , Squared Wasserstein-2 distance for efficient reconstructio n of stochastic differential equations , arXiv preprint arXiv:2401.11354, (2024)

    2024 arXiv

  26. [33]

    Xia and Q

    M. Xia and Q. Shen , A local squared Wasserstein-2 method for efficient reconstru ction of models with uncertainty, arXiv preprint arXiv:2406.06825, (2024)

    2024 arXiv

  27. [35]

    Zheng, F.-Y

    W. Zheng, F.-Y. W ang, and C. Gou , Nonparametric different-feature selection using Wasserste in distance, in ICTAI, IEEE, 2020, pp. 982–988. This manuscript is for review purposes only

    2020

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.