betaFeedback

WorksComputer Science

All roads lead to Markov kernel networks: A representation theorem for discretization-stable learning

Abstract

Operator learning seeks to approximate function-to-function maps in a way that remains stable when the input discretization changes, a property sometimes called "discretization invariance". We consider a quantitative refinement of this idea, called uniform discretization stability\textit{uniform discretization stability}, which builds on a common representation of input functions and their discretizations as probability measures and provides a Hölder bound relating output changes to the Wasserstein distance between these measures. Our main result is a representation theorem showing all uniformly discretization stable operators can be written exactly as a stable Markov kernel network\textit{stable Markov kernel network}—an abstract architecture that is a joint generalization of transformers, Fourier neural operators (FNO), Flowers, and other successful learned operators. Our representation theorem characterizes a design space for discretization-stable operator learning; designing new architectures becomes a question of which parts of a Markov kernel network to parameterize, and how. We further show that the read kernels of these networks can be realized as solutions of regularized variational problems, with particular choices of costs and regularizers recovering the kernels of transformers, FNO, Flowers, their hybrids, and beyond. This variational formulation provides a systematic approach to architecture design, together with sufficient conditions for uniform discretization stability.

Introduction

Operator learning requires neural networks that approximate maps between functions. The input functions of the target operator are continuum objects that cannot be represented exactly on a computer. Furthermore, in scientific applications using real-world data, we generally do not have complete knowledge of the input function ff, but rather only a finite collection of (sometimes noisy) samples from the graph of a function (x,f(x))(x, f(x)). To accurately model the continuum target operator, a neural network should not be sensitive to small changes in the discretization of the input. The output of a “discretization invariant” network converges as the discretization of the input function is refined. Quantitative guarantees about the trained network, however, require stability bounds, not merely convergence under discretization refinement. This leads to the following three questions:

  1. What discretization stability properties are relevant for operator learning?

  2. What network architectures satisfy these stability properties?

  3. How can these architectures be characterized and systematically explored?

Our approach is to lift operators between functions to operators between measures. This provides a unified viewpoint of continuum inputs (functions) and their noisy discretizations. The necessary stability properties can then be stated as a Hölder stability with respect to a Wasserstein metric. Uniform discretization stability controls the behavior of learned operator networks under different discretizations of the input functions. For example, it provides quantitative guarantees for networks that are trained at one resolution or on a uniform grid, but evaluated on non-uniform point clouds sampled at a different resolution. It also provides a unified framework for the analysis of errors arising from both discretization and noisy measurements.

We then turn to the second question, and introduce Markov kernel networks, a shared generalization of transformers (Vaswani et al., 2017), FNO [22], Flowers [26], and other operator networks. These networks alternate nonlinear integral operators built using Markov kernels to mix information spatially with channel-wise mixing operators to mix information pointwise in the embedding space. When the individual components of these networks satisfy relevant stability properties, the entire network becomes uniformly discretization stable. Conversely, our main theorem shows that every uniformly discretization stable operator can be written exactly (not approximately) as a stable Markov kernel network.

With the design space of discretization-stable architectures fixed by the representation theorem, we propose a variational approach to explore the design space. We investigate Markov kernels that minimize regularized objectives; the induced read behavior is determined by the choice of regularizer and cost function. We provide sufficient conditions on the regularizers and costs so that the resulting networks are guaranteed to be uniformly discretization stable, and identify the regularizers corresponding to the popular network architectures discussed earlier. We aim to provide a principled framework for adaptive [6] and hybrid [1] architectures, and a useful belief space for agentic (Toscano et al., 2026) approaches to operator learning.

Prior work

Discretization invariance was initially explored in the context of ill-posed inverse problems. The term was first defined in Lassas & Siltanen [19], with important antecedents in Lasanen [18]; D’Ambrogi et al. [7]; Lehtinen et al. [21]; see Lassas et al. [20] and Stuart [30] for additional historical remarks. It was popularized in operator learning by (earlier preprints of) the seminal work Kovachki et al. [16], which includes analysis of operators that converge under discretization refinement. Various notions of quantitative discretization invariance and stability bounds have been developed in Wang & Golland (2023); Bartolucci et al. [2]; Lanthaler et al. [17]; de Bie et al. [8]; Ruiz et al. [29]; Gao et al. [13]; Subedi & Tewari [31], and Bendahi et al. [3]. Stability of transformers was investigated in Castin et al. [4]. The notion of uniform discretization stability we propose in § 2 has roots in Transportation-LpL^p theory [14, 33]; we are unaware of prior work using these Hölder bounds in the context of operator learning, but Furuya et al. [12] discusses a similar Lipschitz bound for transformers. This paper builds on both the stability bound and “graph measure” viewpoint of that work.

The mindset for our representation theorem and variational characterizations is inspired by Unser (2019), which proved that optimal activation functions (in a total-variation sense) for multi-layer perceptrons take the form of linear splines with adaptive knot locations. Our presentation of Markov kernel networks directly mirrors measure-theoretic transformers [11, 12]. Niculae & Blondel [27] characterized softmax attention as a minimizer of a KL-regularized objective, which was further explored in Martins et al. [25]; Terjék [32].

Paper outline and contributions

  • In § 2, we propose uniform discretization-stability, a Hölder stability property, as a quantitative notion of discretization invariance.

  • In § 3, we define Markov kernel networks, an abstract architecture that generalizes transformers, neural operators, Flowers, and others. Our main results are the representation theorems 3.6 and 3.7 establishing an equivalence between uniformly discretization-stable operators and stable Markov kernel networks. This suggests Markov kernel networks provide a natural design space for uniformly discretization-stable operator learning.

  • § 4 explores the design space of Markov kernel networks using a variational characterization of discretization stable networks. We investigate Markov kernels that minimize a particular regularized objective function, and show how certain regularizers determine different read behavior, including those corresponding to existing architectures and hybrids. We establish stability bounds for variationally defined Markov kernels.

Section 5 discusses some limitations of our results. The appendices contain background and notation used throughout the document (Appendix § 6), proofs (Appendix § 7), a universal approximation result for Markov kernel networks (Appendix § 8), constructions for popular architectures as Markov kernel networks (Appendix § 9), and an additional discussion of hybrid architectures defined by variational Markov reads (Appendix § 10).

Uniform discretization stability

The target operator maps functions to functions, but in practice the network receives samples of the input function and their locations (Figure 1). We represent both these samples and the continuum input as probability measures on a space of functions graphs [12]. The network then maps an input measure to an output function, allowing us to quantify how changes in sampling affect its predictions.

Figure 1. Operator learning setup. The target operator is a continuum object f↦A[f]f \mapsto A[f]. A neural network N\mathcal{N} takes as input a measure μ\mu representing a noisy sample of points from the graph of ff (denoted γf\gamma_f), and outputs a function that can be evaluated at any point in the domain XX. The network N\mathcal{N} is uniformly discretization stable (Definition 2.1) if it satisfies a Hölder stability property that provides quantitative guarantees that N(μ,⋅)\mathcal{N}(\mu,\cdot) converges to N(γf,⋅)\mathcal{N}(\gamma_f,\cdot), the output had the network been provided with the ideal (continuum) input γf\gamma_f.

Let (X,ρX)(X,\rho_X) be a compact metric space with at least two elements, and Y⊂RdoutY \subset\mathbb{R}^{d_{\mathrm{out}}} be non-empty, compact, and convex. Define the graph metric space G:=X×YG := X \times Y equipped with the metric

ρG:G×G→R≥0,ρG((x1,y1),(x2,y2)):=ρX(x1,x2)2+∣y1−y2∣2.(1)\rho_G : G \times G \to\mathbb{R}_{\ge0}, \qquad\rho_G((x_1,y_1),(x_2,y_2)) := \sqrt{\rho_X(x_1,x_2)^2 + \lvert y_1-y_2\rvert^2}. \tag*{(1)}

Choose a Radon probability measure λ∈P(X)\lambda\in\mathcal{P}(X) with supp⁡λ=X\operatorname{supp}\lambda= X. Let L2(X,Y)L^2(X,Y) be the space of Radon-measurable functions f:X→Yf : X \to Y that are square integrable with respect to λ\lambda. Observe that since YY is compact, L2(X,Y)⊂L∞(X,Y)L^2(X,Y) \subset L^\infty(X,Y). In this paper, we consider operators A:L2(X,Y)→L2(X,Y)A : L^2(X,Y) \to L^2(X,Y). All maps in the paper are Borel measurable.

We identify a function f∈L2(X,Y)f \in L^2(X,Y) with its graph measure [12],

γf∈P(G),γf:=(id⁡,f)#λ,(id⁡,f):X→X×Y.(2)\gamma_f \in\mathcal{P}(G), \qquad\gamma_f := (\operatorname{id},f)_\#\lambda,\qquad(\operatorname{id},f) : X \to X \times Y. \tag*{(2)}

(See Appendix § 6 for the definition of the pushforward operator “#\#”.) The map f↦γff \mapsto\gamma_f is injective. For helpful intuition, when ff is continuous (so it is defined pointwise and has a closed graph) we have supp⁡γf={(x,f(x)):x∈X}\operatorname{supp}\gamma_f = \{(x,f(x)) : x \in X\}.

Computer implementations of neural networks for function-to-function maps generally do not accept literal functions f∈L2(X,Y)f \in L^2(X,Y), but instead ff is approximated by a finite collection of points on γf\gamma_f, or only near γf\gamma_f if measurements include noise. We model these sampled points as a mixture of Dirac measures. Numerically, f∈L2(X,Y)f \in L^2(X,Y) is approximated by sampling a finite collection of domain points {xi}i=1N\{x_i\}_{i=1}^{N} and forming an empirical approximation of γf\gamma_f,

μ∈P(G),μ=∑i=1Nwiδ(xi,yi),wi≥0,∑i=1Nwi=1.(3)\mu\in\mathcal{P}(G), \qquad\mu= \sum_{i=1}^{N} w_i\delta_{(x_i,y_i)}, \qquad w_i \ge0,\qquad\sum_{i=1}^{N} w_i = 1. \tag*{(3)}

When the samples (xi,yi)(x_i, y_i) lie on the graph of a function ff, we have yi=f(xi)y_i = f(x_i). The mixture weights wiw_i allow the samples to approximate a graph measure even when the xix_i locations are not distributed uniformly with respect to λ\lambda. This formulation captures both discretizations on uniform grids and irregularly sampled point clouds. Provided with samples μ\mu to represent the input function ff, an operator network N:P(G)×X→Y\mathcal{N} : \mathcal{P}(G) \times X \to Y can then be evaluated at an arbitrary x∈Xx \in X. Informally, a trained neural network N\mathcal{N} is a good approximation of AA when

μ≈γf⟹N(μ,⋅)≈A[f]for all f∈L2(X,Y).(4)\mu\approx\gamma_f \quad\Longrightarrow\quad\mathcal{N}(\mu,\cdot) \approx A[f] \qquad\text{for all } f \in L^2(X,Y). \tag*{(4)}

To make the above precise, we require a metric on measures to compare the graph measure γf\gamma_f to the noisy samples μ\mu. For this, we use the Wasserstein-1 optimal transport metric (Appendix § 6.§ 1).

We now formulate uniform discretization stability as a Hölder-stability property for neural networks.

Definition 2.1 (Uniform discretization stability). A network N:P(G)×X→Y\mathcal{N} : \mathcal{P}(G) \times X \to Y is uniformly discretization stable on a set S⊆P(G)S \subseteq\mathcal{P}(G) if there exist constants C>0C > 0, β∈(0,1]\beta\in(0,1] such that

sup⁡x∈X∣N(μ,x)−N(ν,x)∣≤CW1(μ,ν)βfor all μ,ν∈S.(5)\sup_{x \in X} \left|\mathcal{N}(\mu,x)-\mathcal{N}(\nu,x)\right| \le C W_1(\mu,\nu)^\beta\qquad\text{for all } \mu,\nu\in S. \tag*{(5)}

Returning now to the approximation criterion (4), using the triangle inequality and the fact λ(X)=1\lambda(X)=1,

∥N(μ,⋅)−A[f]∥L2(X,Rdout)≤∥N(γf,⋅)−A[f]∥L2(X,Rdout)+∥N(μ,⋅)−N(γf,⋅)∥∞.(6)\left\|\mathcal{N}(\mu,\cdot)-A[f]\right\|_{L^2(X,\mathbb{R}^{d_{\mathrm{out}}})} \le \left\|\mathcal{N}(\gamma_f,\cdot)-A[f]\right\|_{L^2(X,\mathbb{R}^{d_{\mathrm{out}}})} + \left\|\mathcal{N}(\mu,\cdot)-\mathcal{N}(\gamma_f,\cdot)\right\|_\infty. \tag*{(6)}

The first term on the right is the approximation error of an ideal network given all information about the input function ff; this quantity is controlled by training the network. The second term is the error introduced by the discretization μ\mu. It is largely determined by the neural network architecture itself, and is not directly controlled by training. Discretization stable architectures satisfying (DS) control the discretization error, so they have the possibility to be trained to satisfy the desired property (4).

In the next section, we will show that all such networks have a particular representation as stable Markov kernel networks. This allows us to view, for example, softmax attention, and SelfWarps ([26]) as two particular choices in the same space of discretization stable architectures.

Remark. Uniform discretization stability is a stronger condition than convergence under resolution refinement. In particular, (DS) allows a unified analysis of both resolution refinement and stability to noise in the measurements of ff, both of which are important in applications. Furthermore, for sufficiently regular classes of target operators, it allows us to obtain quantitative guarantees for the performance of a network trained at one resolution but evaluated at another, which is important for scientific applications where large quantities of high-resolution training data may be prohibitively expensive. Indeed, if N\mathcal{N} is uniformly discretization stable, then

∥N(μtest,⋅)−A[f]∥L2(X,Rdout)≤∥N(μtrain,⋅)−A[f]∥L2(X,Rdout)+CW1(μtest,μtrain)β.(7)\left\|\mathcal{N}(\mu_{\mathrm{test}},\cdot)-A[f]\right\|_{L^2(X,\mathbb{R}^{d_{\mathrm{out}}})} \le \left\|\mathcal{N}(\mu_{\mathrm{train}},\cdot)-A[f]\right\|_{L^2(X,\mathbb{R}^{d_{\mathrm{out}}})} + C W_1(\mu_{\mathrm{test}},\mu_{\mathrm{train}})^\beta. \tag*{(7)}

where μtrain\mu_{\mathrm{train}} and μtest\mu_{\mathrm{test}} are discretizations of ff at the training and test resolutions, respectively. Simple convergence under resolution refinement provides no such guarantees.

Remark. Definition 2.1 does not impose additional regularity requirements of the target operator AA or the input functions ff. See the discussion of universal approximation in Appendix § 8.

Non-collapsing measures SS

The set S⊂P(G)S \subset\mathcal{P}(G) in Definition 2.1 need not contain all probability measures to provide a useful notion of discretization stability. Our main result, Theorem 3.7, holds for admissible classes that are relevant for operator learning, but not all possible inputs in P(G)\mathcal{P}(G) (see Example 3.8).

Figure 2. A measure that is not in any admissible set SS. Since (πX)#μ≠μ=δa(\pi_X)_\#\mu\ne\mu= \delta_a, every set containing μ\mu, or a sequence converging to μ\mu, fails the admissibility criterion (7).

A set of measurements at only a single value of xx does not provide much information about an input function ff, and we are not concerned with the behavior of networks on such pathological inputs. To avoid this pathology, we define a condition on admissible sets of measures to ensure there is enough variation in the XX-marginal.

Definition 2.2 (Admissible sets of non-collapsing measures). A non-empty set S⊂P(G)\mathcal{S} \subset\mathcal{P}(\mathcal{G}) is admissible if there exists ϵ>0\epsilon> 0 such that

inf⁡μ∈Sinf⁡a∈XW1((πX)#μ,δa)>ϵ(8)\inf_{\mu\in\mathcal{S}}\inf_{a \in\mathcal{X}} W_1\left((\pi_X)_\#\mu,\delta_a\right) > \epsilon \tag*{(8)}

where πX:(x,y)↦x\pi_X : (x,y) \mapsto x is projection onto the XX component.

Crucially, S\mathcal{S} can contain both graph measures γf\gamma_f and empirical measures μ\mu built from finitely many samples, even in the presence of a large amount of noise, as long as the xx coordinates do not collapse.

Markov kernel networks

We now define an abstract architecture for operator learning called Markov kernel networks, and show they exactly represent all uniformly discretization stable operators on admissible input classes.

Definitions

Let MM be a metric space. A nonlinear Markov kernel is a map K:P1(M)×M→P1(M)K : \mathcal{P}_1(M) \times M \to\mathcal{P}_1(M). Markov kernels form the backbone of the “in-context maps” [11] that define the hidden layers of a Markov kernel network. An in-context map uses a state encoded as a probability measure, called a “context”, to determine how to update a single point.

References

  1. [1]Advait Balaji, Trevor Teolis, S. David Mis, Jose Antonio Lara Benitez, Chao Wang, and Maarten V. de Hoop. Hybrid operator learning of wave scattering maps in high-contrast media, 2026. URL https://arxiv.org/abs/2602.11197v1. Preprint.
  2. [2]Francesca Bartolucci, Emmanuel de Bézenac, Bogdan Raonić, Roberto Molinaro, Siddhartha Mishra, and Rima Alalfiari. Representation equivariant neural operators: a framework for alias-free operator learning. In Advances in Neural Information Processing Systems, volume 36, 2023. URL https://arxiv.org/abs/2305.19913v2.
  3. [3]Abderrahim Bendahi, Adrien Fradin, Johan Peralez, Julie Digne, and Madiha Nadri. Stability and discretization error of state space model neural operators, 2026. URL https://arxiv.org/abs/2605.18905v1.
  4. [4]Valérie Castin, Pierre Ablin, and Gabriel Peyré. How Smooth Is Attention? In Proceedings of the 41st International Conference on Machine Learning, 2024. URL https://arxiv.org/abs/2312.14820v2.
  5. [5]Tianping Chen and Hong Chen. Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems. IEEE Transactions on Neural Networks, 6(4):911–917, 1995. doi: 10.1109/72.392253.DOI
  6. [6]Corinna Cortes, Xavier Gonzalvo, Vitaly Kuznetsov, Mehryar Mohri, and Scott Yang. AdaNet: Adaptive structural learning of artificial neural networks. In Proceedings of the 34th International Conference on Machine Learning, volume 70 of Proceedings of Machine Learning Research, pp. 874–883, 2017. URL https://proceedings.mlr.press/v70/cortes17a.html.arxiv.org/abs/1607.01097
  7. [7]Barbara D’Ambrogi, Sari Mäenpää, and Markku Markkanen. Discretization independent retrieval of atmospheric ozone profile. Geophysica, 35(1-2):87–99, 1999. URL https://archive.geophysica.fi/pdf/geophysica_1999_35_1-2_087_dambrogi.pdf.
  8. [8]Gwendoline de Bie, Gabriel Peyré, and Marco Cuturi. Stochastic deep networks. In Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pp. 1556–1565, 2019. URL https://proceedings.mlr.press/v97/de-biel9a.html.arxiv.org/abs/1811.07429
  9. [9]Miguel de Carvalho, Boris Oumow, Johan Segers, and Michał Warchoł. A Euclidean likelihood estimator for bivariate tail dependence. Communications in Statistics—Theory and Methods, 42(7):1176–1192, 2013. doi: 10.1080/03610926.2012.709905. URL https://arxiv.org/abs/1204.3524.
  10. [11]Takashi Furuya, Maarten V. de Hoop, and Gabriel Peyré. Transformers are universal in-context learners. In The Thirteenth International Conference on Learning Representations, 2025b. URL https://arxiv.org/abs/2408.01367v2. arXiv:2408.01367.
  11. [12]Takashi Furuya, David Mis, Ivan Dokmanić, Maarten V. de Hoop, and Matti Lassas. Function graph transformers universally approximate operators between function spaces, 2026. URL https://arxiv.org/abs/2605.17968v1.
  12. [13]Wenhan Gao, Ruichen Xu, Yuefan Deng, and Yi Liu. Discretization-invariance? on the discretization mismatch errors in neural operators. In International Conference on Learning Representations, 2025. URL https://proceedings.iclr.cc/paper_files/paper/2025/hash/31382957739652c01b5adb3a1cbd9bd-Abstract-Conference.html.
  13. [14]Nicolás García Trillos and Dejan Slepčev. Continuum limit of total variation on point clouds. Archive for Rational Mechanics and Analysis, 220:193–241, 2016. doi: 10.1007/s00205-015-0929-z. URL https://arxiv.org/abs/1403.6355v3.
  14. [15]C. J. Himmelberg and F. S. Van Vleck. Some selection theorems for measurable functions. Canadian Journal of Mathematics, 21:394–399, 1969. doi: 10.4153/CJM-1969-041-7. URL https://www.cambridge.org/core/journals/canadian-journal-of-mathematics/article/some-selection-theorems-for-measurable-functions/303452D43DDFFAAF9CE805C30EE959C6.DOI
  15. [16]Nikola Kovachki, Zongyi Li, Burigede Liu, Kamyar Azizzadenesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural Operator: Learning Maps Between Function Spaces. Journal of Machine Learning Research, 24(89):1–97, 2023. URL https://arxiv.org/abs/2108.08481v6.
  16. [17]Samuel Lanthaler, Andrew M. Stuart, and Margaret Trautner. Discretization error of fourier neural operators, 2024. URL https://arxiv.org/abs/2405.02221. Version 2, 25 September 2025.
  17. [18]Sari Lasanen. Discretizations of Generalized Random Variables with Applications to Inverse Problems. PhD thesis, University of Oulu, Helsinki, 2002. URL https://www.finna.fi/Record/utu.999531305405971. 64 pages.
  18. [19]Matti Lassas and Samuli Siltanen. Can one use total variation prior for edge-preserving Bayesian inversion? Inverse Problems, 20(5):1537–1563, 2004. doi: 10.1088/0266-5611/20/5/013. URL https://siltanen-research.net/wp-content/uploads/publications/LassasSiltanen2004.pdf.DOI
  19. [20]Matti Lassas, Eero Saksman, and Samuli Siltanen. Discretization-invariant Bayesian inversion and Besov space priors. Inverse Problems and Imaging, 3(1):87–122, 2009. doi: 10.3934/ipi.2009.3.87. URL https://www.aimsciences.org/article/doi/10.3934/ipi.2009.3.87.DOI
  20. [21]M. S. Lehtinen, L. Päivärinta, and E. Somersalo. Linear inverse problems for generalised random variables. Inverse Problems, 5(4):599–612, 1989. doi: 10.1088/0266-5611/5/4/011. URL https://doi.org/10.1088/0266-5611/5/4/011.
  21. [22]Zongyi Li, Nikola Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Fourier Neural Operator for Parametric Partial Differential Equations. In International Conference on Learning Representations, 2021. URL https://arxiv.org/abs/2010.08895v3.
  22. [24]André F. T. Martins and Ramón F. Astudillo. From softmax to sparsemax: A sparse model of attention and multi-label classification. In Maria Florina Balcan and Kilian Q. Weinberger (eds.), Proceedings of the 33rd International Conference on Machine Learning, volume 48 of Proceedings of Machine Learning Research, pp. 1614–1623. PMLR, 2016. URL https://proceedings.mlr.press/v48/martins16.html.arxiv.org/abs/1602.02068
  23. [25]André F. T. Martins, Marcos Treviso, António Farinhas, Pedro M. Q. Aguiar, Mário A. T. Figueiredo, Mathieu Blondel, and Vlad Niculae. Sparse continuous distributions and Machine-Learned Young losses. Journal of Machine Learning Research, 23(257):1–74, 2022. URL https://jmlr.org/papers/v23/21-0879.html.
  24. [26]Till Muser, Alexandra Spitzer, Matti Lassas, Maarten V. de Hoop, and Ivan Dokmanić. Flowers: A warp drive for neural PDE solvers. In Forty-Third International Conference on Machine Learning, 2026. URL https://arxiv.org/abs/2603.04430v2. Version 2, 1 June 2026.
  25. [27]Vlad Niculae and Mathieu Blondel. A regularized framework for sparse and structured neural attention. In Advances in Neural Information Processing Systems, volume 30, pp. 3338–3348. Curran Associates, Inc., 2017. URL https://proceedings.neurips.cc/paper/2017/hash/2d1b2a5ff364606ff041650887723470-Abstract.html.arxiv.org/abs/1705.07704
  26. [28]Vincent Roulet, Tianlin Liu, Nino Vieillard, Michaël E. Sander, and Mathieu Blondel. Loss functions and operators generated by f-divergences. In Proceedings of the 42nd International Conference on Machine Learning, volume 267 of Proceedings of Machine Learning Research, pp. 52110–52138. PMLR, 2025. URL https://proceedings.mlr.press/v267/roulet25a.html.arxiv.org/abs/2501.18537
  27. [29]Luana Ruiz, Luiz F. O. Chamon, and Alejandro Ribeiro. Graphon neural networks and the transferability of graph neural networks. In Advances in Neural Information Processing Systems, volume 33, 2020. URL https://papers.nips.cc/paper/2020/hash/12bcd658ef0a540cabc36cdf2b1046fd-Abstract.html.arxiv.org/abs/2006.03548
  28. [30]Andrew M. Stuart. Inverse problems: A Bayesian perspective. Acta Numerica, 19:451–559, 2010. doi: 10.1017/S0962492910000061. URL https://doi.org/10.1017/S0962492910000061.
  29. [31]Unique Subedi and Ambuj Tewari. Controlling statistical, discretization, and truncation errors in learning fourier linear operators, 2025. URL https://arxiv.org/abs/2408.09004. Version 2, 6 February 2025; first posted in 2024 as Error Bounds for Learning Fourier Linear Operators.
  30. [32]Dávid Terjék. Moreau-yosida f-divergences. In Marina Meila and Tong Zhang (eds.), Proceedings of the 38th International Conference on Machine Learning, volume 139 of Proceedings of Machine Learning Research, pp. 10214–10224. PMLR, 2021. URL https://proceedings.mlr.press/v139/terjek21a.html.arxiv.org/abs/2102.13416
  31. [33]Matthew Thorpe, Serim Park, Soheil Kolouri, Gustavo K. Rohde, and Dejan Slepčev. A transportation L^p distance for signal analysis. Journal of Mathematical Imaging and Vision, 59(2):187–210, 2017. doi: 10.1007/s10851-017-0726-4. URL https://pmc.ncbi.nlm.nih.gov/articles/PMC6141213/.DOI

Paper details

vlnc:W_pzjv27gg
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

ChatGPT 6 Astra, Claude

View statement
Looking for related papers…
Contents