mlsanityData to 14 Oct 2026

A Gradient Flow Perspective on Minimum MMD Estimation

arXiv 2607.03871 ↗First posted 4 Jul 2026cs.LG stat.MLcited by 1

Not on the map. Why?

Abstract (arXiv)

Minimum maximum mean discrepancy (MMD) estimation has emerged as a robust and likelihood-free alternative to maximum likelihood estimation for parameter estimation. Yet, despite its practical success, the associated optimization problem remains poorly understood, with theoretical guarantees for existing algorithms hinging on convexity assumptions that rarely hold in practice. We address this gap by proposing a preconditioned gradient descent (PGD) scheme, establishing its asymptotic global convergence under explicit gradient-dominance and projection-residual conditions. Our approach is inspired by recent progress on MMD gradient flows, a nonparametric descent scheme on the space of probability measures. We provide extensive empirical evidence that our PGD scheme outperforms standard gradient descent across a range of challenging parameter estimation and composite hypothesis testing problems.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 57 of its 88 references that are in the map. First posted on arXiv 4 Jul 2026.

Builds on

57 of 88 references.

References · 57

  1. 2016Linear Convergence of Gradient and Proximal-Gradient Methods Under the Polyak-Lojasiewicz Conditioncited by 539
  2. 2014New insights and perspectives on the natural gradient methodcited by 443
  3. 2016Kernel Mean Embedding of Distributions: A Review and Beyondcited by 364
  4. 2013Revisiting Natural Gradient for Deep Networkscited by 159
  5. 2019Maximum Mean Discrepancy Gradient Flowcited by 127
  6. 2017Sobolev Norm Learning Rates for Regularized Least-Squares Algorithmcited by 114
  7. 2020On the Convergence of SGD with Biased Gradientscited by 83
  8. 2019Fast Convergence of Natural Gradient Descent for Overparameterized Neural Networkscited by 76
  9. 2021MMD Aggregated Two-Sample Testcited by 75
  10. 2019Minimum Stein Discrepancy Estimatorscited by 57
  11. 2021Kernel Stein Discrepancy Descentcited by 57
  12. 2022Convex Analysis of the Mean Field Langevin Dynamicscited by 55
38 more
  1. 2019Statistical Inference for Generative Models with Maximum Mean Discrepancycited by 45
  2. 2022Robust Bayesian Inference for Simulator-based Models via the MMD Posterior Bootstrapcited by 41
  3. 2021KALE Flow: A Relaxed KL Gradient Flow for Probabilities with Disjoint Supportcited by 37
  4. 2023Generative Sliced MMD Flows with Riesz Kernelscited by 37
  5. 2020Metrizing Weak Convergence with Maximum Mean Discrepanciescited by 36
  6. 2019MMD-Bayes: Robust Bayesian Estimation via Maximum Mean Discrepancycited by 35
  7. 2015On Graduated Optimization for Stochastic Non-Convex Problemscited by 34
  8. 2023MMD-FUSE: Learning and Combining Kernels for Two-Sample Testing Without Data Splittingcited by 33
  9. 2019Finite sample properties of parametric MMD estimation: robustness to misspecification and dependencecited by 25
  10. 2022Particle-based Variational Inference with Preconditioned Functional Gradient Flowcited by 20
  11. 2024Deep MMD Gradient Flow without adversarial trainingcited by 19
  12. 2021Composite Goodness-of-fit Tests with Kernelscited by 16
  13. 2022Wasserstein Steepest Descent Flows of Discrepancies with Riesz Kernelscited by 16
  14. 2024(De)-regularized Maximum Mean Discrepancy Gradient Flowcited by 16
  15. 2023Optimally-Weighted Estimators of the Maximum Mean Discrepancy for Likelihood-Free Inferencecited by 13
  16. 2025Wasserstein Policy Optimizationcited by 13
  17. 2021Discrepancy-based Inference for Intractable Generative Models using Quasi-Monte Carlocited by 12
  18. 2024Mirror and Preconditioned Gradient Descent in Wasserstein Spacecited by 12
  19. 2025Weighted quantization using MMD: From mean field to mean shift via gradient flowscited by 12
  20. 2026On the Wasserstein Gradient Flow Interpretation of Drifting Modelscited by 11
  21. 2024Conditional Bayesian Quadraturecited by 9
  22. 2019Likelihood-free approximate Gibbs samplingcited by 8
  23. 2020Estimation of copulas via Maximum Mean Discrepancycited by 8
  24. 2024Prediction-Centric Uncertainty Quantification via MMDcited by 8
  25. 2014Characteristic Kernels and Infinitely Divisible Distributionscited by 7
  26. 2023Wasserstein Gradient Flows of the Discrepancy with Distance Kernel on the Linecited by 6
  27. 2026Sobolev Regularized MMD Gradient Flowcited by 6
  28. 2023A High-dimensional Convergence Theorem for U-statistics with Applications to Kernel-based Testingcited by 5
  29. 2025Nonparametric Instrumental Variable Regression with Observed Covariatescited by 5
  30. 2022Minimum Kernel Discrepancy Estimatorscited by 4
  31. 2016Estimating Quantile Families of Loss Distributions for Non-Life Insurance Modelling via L-momentscited by 3
  32. 2021A Deterministic Sampling Method via Maximum Mean Discrepancy Flow with Adaptive Kernelcited by 3
  33. 2023A Semi-Bayesian Nonparametric Estimator of the Maximum Mean Discrepancy Measure: Applications in Goodness-of-Fit Testing and Generative Adversarial Networkscited by 3
  34. 2023Gaussian smoothing gradient descent for minimizing functions (GSmoothGD)cited by 3
  35. 2025Parametric MMD Estimation with Missing Values: Robustness to Missingness and Data Model Misspecificationcited by 3
  36. 2025Total robustness in Bayesian Nonlinear Regressioncited by 3
  37. 2026Minimum Distance Summaries for Robust Neural Posterior Estimationcited by 3
  38. 2021Minimum Wasserstein Distance Estimator under Finite Location-scale Mixturescited by 2

50 most cited of 57 All 57 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026On the Optimisation Landscape of Overparameterised Weighted MMDcited by 0