Efficient Learning of Stationary Diffusions with Stein-type Discrepancies

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bleile, Fabian, Lumpp, Sarah, Drton, Mathias
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911406696169472
author Bleile, Fabian
Lumpp, Sarah
Drton, Mathias
author_facet Bleile, Fabian
Lumpp, Sarah
Drton, Mathias
contents Learning a stationary diffusion amounts to estimating the parameters of a stochastic differential equation whose stationary distribution matches a target distribution. We build on the recently introduced kernel deviation from stationarity (KDS), which enforces stationarity by evaluating expectations of the diffusion's generator in a reproducing kernel Hilbert space. Leveraging the connection between KDS and Stein discrepancies, we introduce the Stein-type KDS (SKDS) as an alternative formulation. We prove that a vanishing SKDS guarantees alignment of the learned diffusion's stationary distribution with the target. Furthermore, under broad parametrizations, SKDS is convex with an empirical version that is $ε$-quasiconvex with high probability. Empirically, learning with SKDS attains comparable accuracy to KDS while substantially reducing computational cost and yields improvements over the majority of competitive baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2601_16597
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Efficient Learning of Stationary Diffusions with Stein-type Discrepancies
Bleile, Fabian
Lumpp, Sarah
Drton, Mathias
Machine Learning
Statistics Theory
60H10, 60J60, 62M99 60H10, 60J60, 62M05
I.2.6; G.3
Learning a stationary diffusion amounts to estimating the parameters of a stochastic differential equation whose stationary distribution matches a target distribution. We build on the recently introduced kernel deviation from stationarity (KDS), which enforces stationarity by evaluating expectations of the diffusion's generator in a reproducing kernel Hilbert space. Leveraging the connection between KDS and Stein discrepancies, we introduce the Stein-type KDS (SKDS) as an alternative formulation. We prove that a vanishing SKDS guarantees alignment of the learned diffusion's stationary distribution with the target. Furthermore, under broad parametrizations, SKDS is convex with an empirical version that is $ε$-quasiconvex with high probability. Empirically, learning with SKDS attains comparable accuracy to KDS while substantially reducing computational cost and yields improvements over the majority of competitive baselines.
title Efficient Learning of Stationary Diffusions with Stein-type Discrepancies
topic Machine Learning
Statistics Theory
60H10, 60J60, 62M99 60H10, 60J60, 62M05
I.2.6; G.3
url https://arxiv.org/abs/2601.16597