Alain Durmus
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Alain Durmus has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Eric Moulines
École Polytechnique
7 shared publications - Sergey V. Samsonov
HSE University, Moscow, Russian Federation
4 shared publications - Alexey Naumov
National Research University Higher School of Economics
3 shared publications - Andreas Eberle
Institute for Applied Mathematics University of Bonn
2 shared publications - Aurélien Enfroy
Département CITI TIPIC Samovar Télécom SudParis Institut Polytechnique de Paris Palaiseau
2 shared publications - Giovanni Conforti
Università degli Studi di Padova Via Trieste 63 35131 Padova Italia
2 shared publications
Latest Publications
- Open Access
Dynamic Programming for Epistemic Uncertainty in Markov Decision Processes
In this paper, we propose a general theory of ambiguity-averse MDPs, which treats the uncertain transition probabilities as random variables and evaluates a policy via a risk measure applied to its random return. This ambiguity-averse MDP framework unifies several models of MDPs with epistemic uncertainty for specific choices of risk measures. We extend the concepts of value functions and Bellman operators to our setting. Based on these objects, we establish the consequences of dynamic programming principles in this framework (existence of stationary policies, value and policy iteration algorithms), and we completely characterize law-invariant risk measures compatible with dynamic programming. Our work draws connections among several variants of MDP models and fully delineates what is possible under the dynamic programming paradigm and which risk measures require leaving it.
Open AccessDynamic Programming for Epistemic Uncertainty in Markov Decision Processes
In this paper, we propose a general theory of ambiguity-averse MDPs, which treats the uncertain transition probabilities as random variables and evaluates a policy via a risk measure applied to its random return. This ambiguity-averse MDP framework unifies several models of MDPs with epistemic uncertainty for specific choices of risk measures. We extend the concepts of value functions and Bellman operators to our setting. Based on these objects, we establish the consequences of dynamic programming principles in this framework (existence of stationary policies, value and policy iteration algorithms), and we completely characterize law-invariant risk measures compatible with dynamic programming. Our work draws connections among several variants of MDP models and fully delineates what is possible under the dynamic programming paradigm and which risk measures require leaving it.
- Open Access
Exponential Convergence Guarantees for Iterative Markovian Fitting
The Schrödinger Bridge (SB) problem has become a fundamental tool in computational optimal transport and generative modeling. To address this problem, ideal methods such as Iterative Proportional Fitting and Iterative Markovian Fitting (IMF) have been proposed-alongside practical approximations like Diffusion Schrödinger Bridge and its Matching (DSBM) variant. While previous work have established asymptotic convergence guarantees for IMF, a quantitative, non-asymptotic understanding remains unknown. In this paper, we provide the first non-asymptotic exponential convergence guarantees for IMF under mild structural assumptions on the reference measure and marginal distributions, assuming a sufficiently large time horizon. Our results encompass two key regimes: one where the marginals are log-concave, and another where they are weakly log-concave. The analysis relies on new contraction results for the Markovian projection operator and paves the way to theoretical guarantees for DSBM.
Open AccessExponential Convergence Guarantees for Iterative Markovian Fitting
The Schrödinger Bridge (SB) problem has become a fundamental tool in computational optimal transport and generative modeling. To address this problem, ideal methods such as Iterative Proportional Fitting and Iterative Markovian Fitting (IMF) have been proposed-alongside practical approximations like Diffusion Schrödinger Bridge and its Matching (DSBM) variant. While previous work have established asymptotic convergence guarantees for IMF, a quantitative, non-asymptotic understanding remains unknown. In this paper, we provide the first non-asymptotic exponential convergence guarantees for IMF under mild structural assumptions on the reference measure and marginal distributions, assuming a sufficiently large time horizon. Our results encompass two key regimes: one where the marginals are log-concave, and another where they are weakly log-concave. The analysis relies on new contraction results for the Markovian projection operator and paves the way to theoretical guarantees for DSBM.
- Open Access
Briding Diffusion Posterior Sampling and Monte Carlo methods: a survey
Diffusion models enable the synthesis of highly accurate samples from complex distributions and have become foundational in generative modeling. Recently, they have demonstrated significant potential for solving Bayesian inverse problems by serving as priors. This review offers a comprehensive overview of current methods that leverage \emph{pre-trained} diffusion models alongside Monte Carlo methods to address Bayesian inverse problems without requiring additional training. We show that these methods primarily employ a \emph{twisting} mechanism for the intermediate distributions within the diffusion process, guiding the simulations toward the posterior distribution. We describe how various Monte Carlo methods are then used to aid in sampling from these twisted distributions.
Open AccessBriding Diffusion Posterior Sampling and Monte Carlo methods: a survey
Diffusion models enable the synthesis of highly accurate samples from complex distributions and have become foundational in generative modeling. Recently, they have demonstrated significant potential for solving Bayesian inverse problems by serving as priors. This review offers a comprehensive overview of current methods that leverage \emph{pre-trained} diffusion models alongside Monte Carlo methods to address Bayesian inverse problems without requiring additional training. We show that these methods primarily employ a \emph{twisting} mechanism for the intermediate distributions within the diffusion process, guiding the simulations toward the posterior distribution. We describe how various Monte Carlo methods are then used to aid in sampling from these twisted distributions.
- Open Access
On the Rate of Gaussian Approximation for Linear Regression Problems
In this paper, we consider the problem of Gaussian approximation for the online linear regression task. We derive the corresponding rates for the setting of a constant learning rate and study the explicit dependence of the convergence rate upon the problem dimension $d$ and quantities related to the design matrix. When the number of iterations $n$ is known in advance, our results yield the rate of normal approximation of order $\sqrt{\log{n}/n}$, provided that the sample size $n$ is large enough.Open AccessOn the Rate of Gaussian Approximation for Linear Regression Problems
In this paper, we consider the problem of Gaussian approximation for the online linear regression task. We derive the corresponding rates for the setting of a constant learning rate and study the explicit dependence of the convergence rate upon the problem dimension $d$ and quantities related to the design matrix. When the number of iterations $n$ is known in advance, our results yield the rate of normal approximation of order $\sqrt{\log{n}/n}$, provided that the sample size $n$ is large enough. - Open Access
- 19 June 2025
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 383 (2299) , 20240334
Generative modelling meets Bayesian inference: a new paradigm for inverse problems
This special issue addresses Bayesian inverse problems using data-driven priors derived from deep generative models (DGMs) and the convergence of generative modelling techniques and Bayesian inference methods. Conventional Bayesian priors often fail to accurately capture the properties and the underlying geometry of complex, real-world data distributions. In contrast, deep generative models (DGMs), which include generative adversarial networks (GANs), variational auto-encoders (VAEs), normalizing flows and diffusion models (DMs), have demonstrated tremendous success in capturing detailed data representations learned directly from empirical observations. As a result, these models produce priors endowed with superior accuracy, increased perceptual realism and enhanced capacities for uncertainty quantification within inverse problem contexts. This paradigm emerged in the late 2010s, when pioneering efforts were made to explicitly formulate Bayesian inverse problems using conditional Wasserstein generative adversarial networks (GANs). These advances have greatly improved methods for quantifying uncertainties, especially in large-scale imaging applications. Building on these fundamental insights, posterior sampling techniques utilizing DMs have demonstrated remarkable efficiency and robustness, highlighting their potential to effectively tackle complex and diverse inverse problems. The articles collected herein provide essential theoretical breakthroughs and significant algorithmic innovations, collectively demonstrating how deep generative priors mitigate traditional limitations and profoundly enrich both the practical applicability and theoretical foundations of Bayesian inversion. This article is part of the theme issue ‘Generative modelling meets Bayesian inference: a new paradigm for inverse problems’.
- 19 June 2025
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 383 (2299) , 20240334
Open AccessGenerative modelling meets Bayesian inference: a new paradigm for inverse problems
This special issue addresses Bayesian inverse problems using data-driven priors derived from deep generative models (DGMs) and the convergence of generative modelling techniques and Bayesian inference methods. Conventional Bayesian priors often fail to accurately capture the properties and the underlying geometry of complex, real-world data distributions. In contrast, deep generative models (DGMs), which include generative adversarial networks (GANs), variational auto-encoders (VAEs), normalizing flows and diffusion models (DMs), have demonstrated tremendous success in capturing detailed data representations learned directly from empirical observations. As a result, these models produce priors endowed with superior accuracy, increased perceptual realism and enhanced capacities for uncertainty quantification within inverse problem contexts. This paradigm emerged in the late 2010s, when pioneering efforts were made to explicitly formulate Bayesian inverse problems using conditional Wasserstein generative adversarial networks (GANs). These advances have greatly improved methods for quantifying uncertainties, especially in large-scale imaging applications. Building on these fundamental insights, posterior sampling techniques utilizing DMs have demonstrated remarkable efficiency and robustness, highlighting their potential to effectively tackle complex and diverse inverse problems. The articles collected herein provide essential theoretical breakthroughs and significant algorithmic innovations, collectively demonstrating how deep generative priors mitigate traditional limitations and profoundly enrich both the practical applicability and theoretical foundations of Bayesian inversion. This article is part of the theme issue ‘Generative modelling meets Bayesian inference: a new paradigm for inverse problems’.
- 20 February 2025
- Published by Springer Nature
- p. 431-450
Integrating FedDRL for Efficient Vehicular Communication in Smart Cities
- 20 February 2025
- Published by Springer Nature
- p. 431-450
Integrating FedDRL for Efficient Vehicular Communication in Smart Cities
- Open Access
Theoretical guarantees in KL for Diffusion Flow Matching
Flow Matching (FM) (also referred to as stochastic interpolants or rectified flows) stands out as a class of generative models that aims to bridge in finite time the target distribution $ν^\star$ with an auxiliary distribution $μ$, leveraging a fixed coupling $π$ and a bridge which can either be deterministic or stochastic. These two ingredients define a path measure which can then be approximated by learning the drift of its Markovian projection. The main contribution of this paper is to provide relatively mild assumptions on $ν^\star$, $μ$ and $π$ to obtain non-asymptotics guarantees for Diffusion Flow Matching (DFM) models using as bridge the conditional distribution associated with the Brownian motion. More precisely, we establish bounds on the Kullback-Leibler divergence between the target distribution and the one generated by such DFM models under moment conditions on the score of $ν^\star$, $μ$ and $π$, and a standard $L^2$-drift-approximation error assumption.Open AccessTheoretical guarantees in KL for Diffusion Flow Matching
Flow Matching (FM) (also referred to as stochastic interpolants or rectified flows) stands out as a class of generative models that aims to bridge in finite time the target distribution $ν^\star$ with an auxiliary distribution $μ$, leveraging a fixed coupling $π$ and a bridge which can either be deterministic or stochastic. These two ingredients define a path measure which can then be approximated by learning the drift of its Markovian projection. The main contribution of this paper is to provide relatively mild assumptions on $ν^\star$, $μ$ and $π$ to obtain non-asymptotics guarantees for Diffusion Flow Matching (DFM) models using as bridge the conditional distribution associated with the Brownian motion. More precisely, we establish bounds on the Kullback-Leibler divergence between the target distribution and the one generated by such DFM models under moment conditions on the score of $ν^\star$, $μ$ and $π$, and a standard $L^2$-drift-approximation error assumption. - Open Access
On Sampling with Approximate Transport Maps
Transport maps can ease the sampling of distributions with non-trivial geometries by transforming them into distributions that are easier to handle. The potential of this approach has risen with the development of Normalizing Flows (NF) which are maps parameterized with deep neural networks trained to push a reference distribution towards a target. NF-enhanced samplers recently proposed blend (Markov chain) Monte Carlo methods with either (i) proposal draws from the flow or (ii) a flow-based reparametrization. In both cases, the quality of the learned transport conditions performance. The present work clarifies for the first time the relative strengths and weaknesses of these two approaches. Our study concludes that multimodal targets can be reliably handled with flow-based proposals up to moderately high dimensions. In contrast, methods relying on reparametrization struggle with multimodality but are more robust otherwise in high-dimensional settings and under poor training. To further illustrate the influence of target-proposal adequacy, we also derive a new quantitative bound for the mixing time of the Independent Metropolis-Hastings sampler.
Open AccessOn Sampling with Approximate Transport Maps
Transport maps can ease the sampling of distributions with non-trivial geometries by transforming them into distributions that are easier to handle. The potential of this approach has risen with the development of Normalizing Flows (NF) which are maps parameterized with deep neural networks trained to push a reference distribution towards a target. NF-enhanced samplers recently proposed blend (Markov chain) Monte Carlo methods with either (i) proposal draws from the flow or (ii) a flow-based reparametrization. In both cases, the quality of the learned transport conditions performance. The present work clarifies for the first time the relative strengths and weaknesses of these two approaches. Our study concludes that multimodal targets can be reliably handled with flow-based proposals up to moderately high dimensions. In contrast, methods relying on reparametrization struggle with multimodality but are more robust otherwise in high-dimensional settings and under poor training. To further illustrate the influence of target-proposal adequacy, we also derive a new quantitative bound for the mixing time of the Independent Metropolis-Hastings sampler.
- 18 February 2024
- Published by Springer Nature in Journal of Theoretical Probability
- Vol. 37 (3) , 2184-2233
Probability and Moment Inequalities for Additive Functionals of Geometrically Ergodic Markov Chains
- 18 February 2024
- Published by Springer Nature in Journal of Theoretical Probability
- Vol. 37 (3) , 2184-2233
Probability and Moment Inequalities for Additive Functionals of Geometrically Ergodic Markov Chains
- Open Access
- 6 November 2023
- Published by Springer Nature in Stochastics and Partial Differential Equations: Analysis and Computations
- Vol. 12 (3) , 1855-1906
Sticky nonlinear SDEs and convergence of McKean–Vlasov equations without confinement
We develop a new approach to study the long time behaviour of solutions to nonlinear stochastic differential equations in the sense of McKean, as well as propagation of chaos for the corresponding mean-field particle system approximations. Our approach is based on a sticky coupling between two solutions to the equation. We show that the distance process between the two copies is dominated by a solution to a one-dimensional nonlinear stochastic differential equation with a sticky boundary at zero. This new class of equations is then analyzed carefully. In particular, we show that the dominating equation has a phase transition. In the regime where the Dirac measure at zero is the only invariant probability measure, we prove exponential convergence to equilibrium both for the one-dimensional equation, and for the original nonlinear SDE. Similarly, propagation of chaos is shown by a componentwise sticky coupling and comparison with a system of one dimensional nonlinear SDEs with sticky boundaries at zero. The approach applies to equations without confinement potential and to interaction terms that are not of gradient type.
- 6 November 2023
- Published by Springer Nature in Stochastics and Partial Differential Equations: Analysis and Computations
- Vol. 12 (3) , 1855-1906
Open AccessSticky nonlinear SDEs and convergence of McKean–Vlasov equations without confinement
We develop a new approach to study the long time behaviour of solutions to nonlinear stochastic differential equations in the sense of McKean, as well as propagation of chaos for the corresponding mean-field particle system approximations. Our approach is based on a sticky coupling between two solutions to the equation. We show that the distance process between the two copies is dominated by a solution to a one-dimensional nonlinear stochastic differential equation with a sticky boundary at zero. This new class of equations is then analyzed carefully. In particular, we show that the dominating equation has a phase transition. In the regime where the Dirac measure at zero is the only invariant probability measure, we prove exponential convergence to equilibrium both for the one-dimensional equation, and for the original nonlinear SDE. Similarly, propagation of chaos is shown by a componentwise sticky coupling and comparison with a system of one dimensional nonlinear SDEs with sticky boundaries at zero. The approach applies to equations without confinement potential and to interaction terms that are not of gradient type.