pith. sign in

arxiv: 2606.02047 · v1 · pith:TDJYRXKGnew · submitted 2026-06-01 · 📊 stat.ML · cs.LG· math.ST· stat.ME· stat.TH

Convex Distance Operator Transport: A Convex and Geometry-Preserving Formulation

Junhyoung Chung , Euijong Song , Won Hwa Kim , Gunwoong Park This is my paper

Pith reviewed 2026-06-28 12:54 UTC · model grok-4.3

classification 📊 stat.ML cs.LGmath.STstat.MEstat.TH
keywords optimal transportconvex optimizationdistribution alignmentgeometric structurepseudometricGromov-WassersteinFrank-Wolfeheterogeneous domains
0
0 comments X

The pith

CDOT is a convex optimal transport method that aligns distributions from different domains while preserving both feature correspondence and intrinsic geometric structure.

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

The paper presents Convex Distance Operator Transport as the first convex framework for optimal transport across heterogeneous domains. It introduces an operator-based regularization using distance and conditional expectation operators to align aggregated distance structures. This approach yields a valid pseudometric on attributed compact metric-measure spaces and provides non-asymptotic risk bounds that establish consistency under the Frank-Wolfe algorithm. The work also relates CDOT to Gromov-Wasserstein by introducing a dispersion gap to identify the geometric origin of non-convexity in the latter. Experiments on point clouds, connectomes, and graphs show improved performance and stability compared to prior methods.

Core claim

CDOT is the first convex optimal transport framework that aligns distributions across heterogeneous domains by jointly preserving feature correspondence and intrinsic geometric structure; the resulting discrepancy is a valid pseudometric on attributed compact metric-measure spaces, with non-asymptotic risk consistency under Frank-Wolfe.

What carries the argument

Operator-based regularization with distance and conditional expectation operators that aligns aggregated distance structures to enforce convexity while improving robustness to local geometric variations.

Load-bearing premise

The distance and conditional expectation operators align aggregated distance structures sufficiently to guarantee both convexity and pseudometric properties without introducing hidden non-convexity.

What would settle it

A pair of attributed compact metric-measure spaces where the CDOT value violates the triangle inequality, or where Frank-Wolfe optimization fails to reach a global minimum with the claimed risk bound, would falsify the central claims.

Figures

Figures reproduced from arXiv: 2606.02047 by Euijong Song, Gunwoong Park, Junhyoung Chung, Won Hwa Kim.

Figure 1
Figure 1. Figure 1: Cyclic graphs with different numbers of nodes. While FGW compares distances edge by edge (|d − d ′ | 2 ), CDOT aligns the aggregated distance profiles (|E[d] − E[d ′ ]| 2 ), comparing how each node relates to its geometric structure. 2008). As OT induces meaningful correspondences between samples, it has become a powerful tool in statistics and ma￾chine learning with various applications ranging from gen￾e… view at source ↗
Figure 2
Figure 2. Figure 2: Comparison of level sets. The blue region represents the range of π 7→ (R(π), V(π)), and red lines indicate the level sets of the respective objectives. πind := PX ⊗ PY denotes the independent coupling. (a) Our formulation yields vertical level sets, guiding the optimization trajectory (black arrows) from initialization π0 directly toward the global optimum π ∗ . (b) In contrast, the GW objective induces d… view at source ↗
Figure 3
Figure 3. Figure 3: Visualization of an OASIS-3 brain connectome (170 nodes). Nodes are colored according to six anatomical region labels, corresponding to left, right, and central regions, each subdi￾vided into cortical and subcortical areas. consists of 170 nodes annotated by six anatomical labels (see [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Empirical visualization of the optimization landscape. The scatter plot illustrates the image of the transport polytope projected onto the plane spanned by the structural regularization R(π) (x-axis) and the dispersion penalty V(π) (y-axis) for an isomorphic matching problem with n = 4. Black circles represent permutation matrices (extreme points), the red triangle marks the independent coupling (PX ⊗ PY )… view at source ↗
Figure 5
Figure 5. Figure 5: Proof roadmap for the four main theoretical results of CDOT. The orange banner names the foundational tool (the disintegration theorem), and the four panels list the technical ingredients feeding into each of the navy theorem boxes. E. Proof Roadmap This appendix collects the technical machinery that supports our four main theoretical results. Before diving into the lemmas, [PITH_FULL_IMAGE:figures/full_f… view at source ↗
Figure 6
Figure 6. Figure 6: Voronoi tessellations generated from samples drawn from the uniform distribution U([0, 1]2 ). F.3.1. DUAL FORM OF THE WASSERSTEIN-1 DISTANCE Definition F.6 (Wasserstein-p distance). Let (S, dS ) be a compact metric space. For Borel probability measures µ and ν on S, the Wasserstein-p distance (1 ≤ p < ∞) is WdS p (µ, ν) :=  min π∈Π(µ,ν) Z S×S dS (x, y) p π(dx, dy) 1/p . We recall the definition of the Wa… view at source ↗
read the original abstract

We introduce Convex Distance Operator Transport (CDOT), the first convex optimal transport framework that aligns distributions across heterogeneous domains by jointly preserving feature correspondence and intrinsic geometric structure. Specifically, CDOT employs an operator-based regularization that aligns aggregated distance structures by introducing distance and conditional expectation operators. Consequently, the proposed regularization improves the robustness to local geometric variations. We further prove that the resulting CDOT discrepancy is a valid pseudometric on the space of attributed compact metric-measure spaces. In addition, we characterize the relationship between CDOT and Gromov--Wasserstein (GW) through a new notion of dispersion gap, formally elucidating the geometric source of non-convexity in GW compared to the convexity of CDOT. In the finite-sample regime, we derive a non-asymptotic risk bound decomposed into optimization and statistical errors, establishing risk consistency under a globally convergent Frank--Wolfe algorithm. Experiments on synthetic point clouds, brain connectomes, and graph classification benchmarks demonstrate better performance over existing methods, with stable and reliable behavior in practice.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper introduces Convex Distance Operator Transport (CDOT), a convex optimal transport framework for aligning distributions across heterogeneous domains. It uses distance and conditional expectation operators to regularize the objective, jointly preserving feature correspondence and intrinsic geometry. The manuscript claims to prove that the resulting discrepancy is a valid pseudometric on attributed compact metric-measure spaces, introduces a dispersion-gap notion to relate CDOT to Gromov-Wasserstein (GW) and explain GW's non-convexity, derives non-asymptotic risk bounds decomposed into optimization and statistical errors, and establishes consistency under a globally convergent Frank-Wolfe algorithm. Experiments on synthetic point clouds, brain connectomes, and graph classification tasks are reported to show improved performance and stability.

Significance. If the convexity of the objective and the pseudometric property hold with the stated proofs, CDOT would provide a theoretically grounded convex alternative to non-convex GW distances, enabling reliable optimization while retaining geometric fidelity. The dispersion-gap characterization and non-asymptotic bounds under Frank-Wolfe would be notable contributions for understanding and mitigating non-convexity in geometric OT. The empirical results on heterogeneous data would support practical utility in domain alignment tasks.

major comments (3)
  1. [Abstract / CDOT objective definition] Abstract and the section defining the CDOT objective: the central convexity claim rests on the assertion that the distance and conditional-expectation operators yield a jointly convex functional in the coupling π. However, conditional expectation is measure-dependent; without an explicit expansion of the regularized objective as a function of π (showing linearity or convexity preservation in each term), it remains possible that the marginal dependence introduces a non-convex term. This is load-bearing for the claim that CDOT is convex while GW is not.
  2. [Pseudometric proof section] Section proving the pseudometric property: the proof that CDOT satisfies the pseudometric axioms (non-negativity, symmetry, triangle inequality) on attributed compact metric-measure spaces must be verified against the operator definitions. If the operators are defined in a manner that depends on the marginals in a non-linear way, the triangle inequality may fail to hold in the claimed form; the manuscript should supply the full derivation rather than asserting the result.
  3. [Risk bound theorem] Section on non-asymptotic risk bounds: the decomposition into optimization and statistical errors and the consistency result under Frank-Wolfe assumes global convergence of the algorithm on the CDOT objective. The bound's dependence on the dispersion gap and the operator norms should be stated explicitly (e.g., via the relevant theorem number); without the explicit assumptions on the operators, it is unclear whether the statistical error term remains controlled when the conditional expectation couples back to the empirical measures.
minor comments (2)
  1. [Notation / operator definitions] Notation for the distance operator and conditional expectation operator should be introduced with explicit functional forms (e.g., as maps from measures to operators) before their use in the objective; current presentation leaves their precise action on the coupling implicit.
  2. [Experiments] The experimental section would benefit from reporting the specific values of any hyperparameters controlling the operator regularization strength, together with sensitivity analysis, to allow reproduction of the reported stability advantages.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments. We address each major point below and will revise the manuscript to incorporate explicit expansions and full derivations as requested.

read point-by-point responses

  1. Referee: [Abstract / CDOT objective definition] Abstract and the section defining the CDOT objective: the central convexity claim rests on the assertion that the distance and conditional-expectation operators yield a jointly convex functional in the coupling π. However, conditional expectation is measure-dependent; without an explicit expansion of the regularized objective as a function of π (showing linearity or convexity preservation in each term), it remains possible that the marginal dependence introduces a non-convex term. This is load-bearing for the claim that CDOT is convex while GW is not.

    Authors: We agree that an explicit expansion of the objective as a function of π would strengthen the convexity claim. In the revision we will provide the full expansion, showing that the distance-operator term is linear in π while the conditional-expectation term, under the operator definition on compact metric-measure spaces, preserves joint convexity in the coupling. revision: yes

  2. Referee: [Pseudometric proof section] Section proving the pseudometric property: the proof that CDOT satisfies the pseudometric axioms (non-negativity, symmetry, triangle inequality) on attributed compact metric-measure spaces must be verified against the operator definitions. If the operators are defined in a manner that depends on the marginals in a non-linear way, the triangle inequality may fail to hold in the claimed form; the manuscript should supply the full derivation rather than asserting the result.

    Authors: We will include the complete step-by-step derivation of all pseudometric axioms in the revised main text (moving the appendix sketch forward), explicitly verifying each step against the operator definitions and confirming that the triangle inequality holds via the metric properties of the distance and conditional-expectation operators. revision: yes

  3. Referee: [Risk bound theorem] Section on non-asymptotic risk bounds: the decomposition into optimization and statistical errors and the consistency result under Frank-Wolfe assumes global convergence of the algorithm on the CDOT objective. The bound's dependence on the dispersion gap and the operator norms should be stated explicitly (e.g., via the relevant theorem number); without the explicit assumptions on the operators, it is unclear whether the statistical error term remains controlled when the conditional expectation couples back to the empirical measures.

    Authors: We will revise the risk-bound section to state the explicit dependence on the dispersion gap and operator norms (with theorem references), clarify the operator assumptions, and demonstrate control of the statistical error term under the empirical measures given the global convergence of Frank-Wolfe. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation self-contained with independent proofs

full rationale

The abstract and provided text introduce CDOT via operator regularization, prove it is a pseudometric on attributed compact metric-measure spaces, characterize its relation to GW via a new dispersion gap notion, and derive non-asymptotic risk bounds under Frank-Wolfe. No equations, definitions, or steps reduce by construction to fitted inputs, self-definitions, or self-citation chains. The central convexity and pseudometric claims are presented as derived from the operator construction and standard measure-theoretic arguments rather than presupposing the target result. This is the normal case of a self-contained theoretical contribution.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are detailed beyond standard metric-measure space assumptions.

axioms (1)
  • domain assumption Attributed compact metric-measure spaces form the underlying domain on which the pseudometric is defined.
    Invoked when stating that CDOT is a valid pseudometric on this space.
invented entities (1)
  • Distance operator and conditional expectation operator no independent evidence
    purpose: Regularize the transport plan by aligning aggregated distance structures
    New operators introduced in the CDOT formulation; no independent evidence supplied in abstract.

pith-pipeline@v0.9.1-grok · 5726 in / 1384 out tokens · 34112 ms · 2026-06-28T12:54:47.499112+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

79 extracted references

  1. [1]

    Communications on pure and applied mathematics , volume=

    Polar factorization and monotone rearrangement of vector-valued functions , author=. Communications on pure and applied mathematics , volume=. 1991 , publisher=

    1991

  2. [2]

    Vayer, Titouan and Chapel, Laetitia and Flamary, R. Fused. Algorithms , volume=. 2020 , publisher=

    2020

  3. [3]

    2008 , publisher=

    Optimal transport: old and new , author=. 2008 , publisher=

    2008

  4. [4]

    Foundations and Trends

    Computational optimal transport: With applications to data science , author=. Foundations and Trends. 2019 , publisher=

    2019

  5. [5]

    Convergence of a

    Kitagawa, Jun and M. Convergence of a. Journal of the European Mathematical Society , volume=

  6. [6]

    2015 , publisher=

    Optimal transport for applied mathematicians , author=. 2015 , publisher=

    2015

  7. [7]

    Sharp asymptotic and finite-sample rates of convergence of empirical measures in

    Weed, Jonathan and Bach, Francis , journal=. Sharp asymptotic and finite-sample rates of convergence of empirical measures in. 2019 , publisher=

    2019

  8. [8]

    On the rate of convergence in

    Fournier, Nicolas and Guillin, Arnaud , journal=. On the rate of convergence in. 2015 , publisher=

    2015

  9. [9]

    Cognitive Neurodynamics , volume=

    The braingraph.org database of high resolution structural connectomes and the brain graph tools , author=. Cognitive Neurodynamics , volume=. 2017 , publisher=

    2017

  10. [10]

    1998 , publisher=

    Mass Transportation Problems: Volume I: Theory , author=. 1998 , publisher=

    1998

  11. [11]

    2021 , publisher=

    Topics in optimal transportation , author=. 2021 , publisher=

    2021

  12. [12]

    International journal of computer vision , volume=

    The earth mover's distance as a metric for image retrieval , author=. International journal of computer vision , volume=. 2000 , publisher=

    2000

  13. [13]

    IEEE signal processing magazine , volume=

    Optimal mass transport: Signal processing and machine-learning applications , author=. IEEE signal processing magazine , volume=. 2017 , publisher=

    2017

  14. [14]

    Statistical aspects of

    Panaretos, Victor M and Zemel, Yoav , journal=. Statistical aspects of. 2019 , publisher=

    2019

  15. [15]

    Advances in Neural Information Processing Systems , volume=

    Deep shells: Unsupervised shape correspondence with optimal transport , author=. Advances in Neural Information Processing Systems , volume=

  16. [16]

    IEEE transactions on pattern analysis and machine intelligence , volume=

    Optimal transport for domain adaptation , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2016 , publisher=

    2016

  17. [17]

    Joint European conference on machine learning and knowledge discovery in databases , pages=

    Domain adaptation with regularized optimal transport , author=. Joint European conference on machine learning and knowledge discovery in databases , pages=. 2014 , organization=

    2014

  18. [18]

    Proceedings of the European conference on computer vision (ECCV) , pages=

    Damodaran, Bharath Bhushan and Kellenberger, Benjamin and Flamary, R. Proceedings of the European conference on computer vision (ECCV) , pages=

  19. [19]

    Foundations of computational mathematics , volume=

    M. Foundations of computational mathematics , volume=. 2011 , publisher=

    2011

  20. [20]

    International Conference on Machine Learning , pages=

    Peyr. International Conference on Machine Learning , pages=. 2016 , organization=

    2016

  21. [21]

    Semidefinite relaxations of the

    Chen, Junyu and Nguyen, Binh T and Koh, Shang and Soh, Yong Sheng , journal=. Semidefinite relaxations of the

  22. [22]

    Entropic

    Rioux, Gabriel and Goldfeld, Ziv and Kato, Kengo , journal=. Entropic

  23. [23]

    Journal of Machine Learning Research , volume=

    Flamary, R. Journal of Machine Learning Research , volume=

  24. [24]

    1993 , publisher=

    Network Flows: Theory, Algorithms, and Applications , author=. 1993 , publisher=

    1993

  25. [25]

    Cuturi, Marco , booktitle=

  26. [26]

    Near-linear time approximation algorithms for optimal transport via

    Altschuler, Jason and Weed, Jonathan and Rigollet, Philippe , booktitle=. Near-linear time approximation algorithms for optimal transport via

  27. [27]

    , journal=

    Kuhn, Harold W. , journal=. The

  28. [28]

    Computing , volume=

    A shortest augmenting path algorithm for dense and sparse linear assignment problems , author=. Computing , volume=

  29. [29]

    Acta Mathematica , volume=

    On the geometry of metric measure spaces , author=. Acta Mathematica , volume=. 2006 , publisher=

    2006

  30. [30]

    Proceedings of the National Academy of Sciences , volume=

    Generalized multidimensional scaling: a framework for isometry-invariant partial surface matching , author=. Proceedings of the National Academy of Sciences , volume=. 2006 , publisher=

    2006

  31. [31]

    ACM Transactions on Graphics (ToG) , volume=

    Functional maps: a flexible representation of maps between shapes , author=. ACM Transactions on Graphics (ToG) , volume=. 2012 , publisher=

    2012

  32. [32]

    Chowdhury, Samir and Needham, Tom , booktitle=

  33. [33]

    Naval Research Logistics Quarterly , volume=

    An algorithm for quadratic programming , author=. Naval Research Logistics Quarterly , volume=. 1956 , publisher=

    1956

  34. [34]

    USSR Computational mathematics and mathematical physics , volume=

    Constrained minimization methods , author=. USSR Computational mathematics and mathematical physics , volume=. 1966 , publisher=

    1966

  35. [35]

    Revisiting

    Jaggi, Martin , booktitle=. Revisiting. 2013 , organization=

    2013

  36. [36]

    2004 , publisher=

    Convex optimization , author=. 2004 , publisher=

    2004

  37. [37]

    2013 , publisher=

    Introductory lectures on convex optimization: A basic course , author=. 2013 , publisher=

    2013

  38. [38]

    Foundations and Trends

    Convex optimization: Algorithms and complexity , author=. Foundations and Trends. 2015 , publisher=

    2015

  39. [39]

    Boosting

    Combettes, Cyrille and Pokutta, Sebastian , booktitle=. Boosting. 2020 , organization=

    2020

  40. [40]

    Operations Research Letters , volume=

    Complexity of linear minimization and projection on some sets , author=. Operations Research Letters , volume=. 2021 , publisher=

    2021

  41. [41]

    2021 , publisher=

    Bomze, Immanuel M and Rinaldi, Francesco and Zeffiro, Damiano , journal=. 2021 , publisher=

    2021

  42. [42]

    A tight upper bound on the rate of convergence of

    Canon, Michael D and Cullum, Clifton D , journal=. A tight upper bound on the rate of convergence of. 1968 , publisher=

    1968

  43. [43]

    ACM Transactions on Graphics (ToG) , volume=

    Entropic metric alignment for correspondence problems , author=. ACM Transactions on Graphics (ToG) , volume=. 2016 , publisher=

    2016

  44. [44]

    Titouan, Vayer and Flamary, R. Sliced. Advances in Neural Information Processing Systems , volume=

  45. [45]

    The unbalanced

    S. The unbalanced. Advances in Neural Information Processing Systems , volume=

  46. [46]

    Semi-relaxed

    Vincent-Cuaz, C. Semi-relaxed. Colloque GRETSI 2022-XXVIII

    2022

  47. [47]

    Proceedings of the National Academy of Sciences , volume=

    On convex relaxation of graph isomorphism , author=. Proceedings of the National Academy of Sciences , volume=. 2015 , publisher=

    2015

  48. [48]

    IEEE transactions on pattern analysis and machine intelligence , volume=

    A path following algorithm for the graph matching problem , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2008 , publisher=

    2008

  49. [49]

    Asian Conference on Machine Learning , pages=

    A convex-concave relaxation procedure based subgraph matching algorithm , author=. Asian Conference on Machine Learning , pages=. 2012 , organization=

    2012

  50. [50]

    Quadratic programming relaxations for metric labeling and

    Ravikumar, Pradeep and Lafferty, John , booktitle=. Quadratic programming relaxations for metric labeling and

  51. [51]

    Image and Vision Computing , volume=

    Evaluation of a convex relaxation to a quadratic assignment matching approach for relational object views , author=. Image and Vision Computing , volume=. 2007 , publisher=

    2007

  52. [52]

    International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition , pages=

    Probabilistic subgraph matching based on convex relaxation , author=. International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition , pages=. 2005 , organization=

    2005

  53. [53]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    A convex relaxation for multi-graph matching , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  54. [54]

    Stability of optimal transport maps on

    Kitagawa, Jun and Letrouit, Cyril and M. Stability of optimal transport maps on. arXiv preprint arXiv:2504.05412 , year=

    arXiv

  55. [55]

    Polar factorization of maps on

    McCann, Robert J , journal=. Polar factorization of maps on. 2001 , publisher=

    2001

  56. [56]

    Volume II: Advanced theory , author=

    Fundamentals of the theory of operator algebras. Volume II: Advanced theory , author=. 1986 , publisher=

    1986

  57. [57]

    Real analysis: measure theory, integration, and

    Stein, Elias M and Shakarchi, Rami , year=. Real analysis: measure theory, integration, and

  58. [58]

    2019 , publisher=

    A course in functional analysis , author=. 2019 , publisher=

    2019

  59. [59]

    2006 , publisher=

    Infinite dimensional analysis: a hitchhiker’s guide , author=. 2006 , publisher=

    2006

  60. [60]

    Wu, Zhirong and Song, Shuran and Khosla, Aditya and Yu, Fisher and Zhang, Linguang and Tang, Xiaoou and Xiao, Jianxiong , booktitle=

  61. [61]

    Applied and computational harmonic analysis , volume=

    Diffusion maps , author=. Applied and computational harmonic analysis , volume=. 2006 , publisher=

    2006

  62. [62]

    Neuron , volume=

    Cooperative and competitive spreading dynamics on the human connectome , author=. Neuron , volume=. 2015 , publisher=

    2015

  63. [63]

    Neuroimage , volume=

    Network diffusion accurately models the relationship between structural and functional brain connectivity networks , author=. Neuroimage , volume=. 2014 , publisher=

    2014

  64. [64]

    Advances in Neural Information Processing Systems , volume=

    Co-optimal transport , author=. Advances in Neural Information Processing Systems , volume=

  65. [65]

    Dong, Yihe and Sawin, Will , journal=

  66. [66]

    A novel sliced fused

    Piening, Moritz and Beinert, Robert , journal=. A novel sliced fused

  67. [67]

    Linear-time

    Scetbon, Meyer and Peyr. Linear-time. International Conference on Machine Learning , pages=. 2022 , organization=

    2022

  68. [68]

    Orthogonal

    Jin, Hongwei and Yu, Zishun and Zhang, Xinhua , booktitle=. Orthogonal. 2022 , organization=

    2022

  69. [69]

    Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1 , volume=

    A spectral technique for correspondence problems using pairwise constraints , author=. Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1 , volume=. 2005 , organization=

    2005

  70. [70]

    Proceedings of the National Academy of Sciences , volume=

    Global alignment of multiple protein interaction networks with application to functional orthology detection , author=. Proceedings of the National Academy of Sciences , volume=. 2008 , publisher=

    2008

  71. [71]

    Econometrica: journal of the Econometric Society , pages=

    Assignment problems and the location of economic activities , author=. Econometrica: journal of the Econometric Society , pages=. 1957 , publisher=

    1957

  72. [72]

    Kriege and Franka Bause and Kristian Kersting and Petra Mutzel and Marion Neumann , booktitle=

    Christopher Morris and Nils M. Kriege and Franka Bause and Kristian Kersting and Petra Mutzel and Marion Neumann , booktitle=

  73. [73]

    Journal of the American Statistical Association , volume=

    Optimal Transport based Cross-Domain Integration for Heterogeneous Data , author=. Journal of the American Statistical Association , volume=. 2025 , publisher=

    2025

  74. [74]

    The Annals of Statistics , volume=

    Plugin estimation of smooth optimal transport maps , author=. The Annals of Statistics , volume=. 2024 , publisher=

    2024

  75. [75]

    The Annals of Statistics , volume=

    Minimax estimation of smooth optimal transport maps , author=. The Annals of Statistics , volume=. 2021 , publisher=

    2021

  76. [76]

    The Annals of Statistics , volume=

    Optimal transport map estimation in general function spaces , author=. The Annals of Statistics , volume=. 2025 , publisher=

    2025

  77. [77]

    Advances in Neural Information Processing Systems , volume=

    Rates of estimation of optimal transport maps using plug-in estimators via barycentric projections , author=. Advances in Neural Information Processing Systems , volume=

  78. [78]

    Combinatorica , volume=

    Fractional isomorphism of graphons , author=. Combinatorica , volume=. 2022 , publisher=

    2022

  79. [79]

    The Annals of Applied Probability , volume=

    Stability of martingale optimal transport and weak optimal transport , author=. The Annals of Applied Probability , volume=. 2022 , publisher=

    2022