The paper introduces β‑Flow, a versatile plug‑and‑play posterior sampling framework that unifies flow‑matching, diffusion, and decorrelated priors through a correlation parameter β and a confidence parameter κ. Each iteration requires only a single denoising network forward pass followed by an exact Gaussian conditioning step, implemented as a Kalman update independent of network gradients. The authors demonstrate that many existing guided samplers are special cases of β‑Flow, and show state‑of‑the‑art performance on seven inverse problems across four datasets, with the ability to tune prior strength based on data informativeness.
arXiv:2606. 17048v1 Announce Type: new Abstract: Diffusion and flow-based models learn powerful data priors by training a denoiser to reverse Gaussian corruption.
The paper introduces Latent Annealed Posterior Sampling (LAPS), a method that maps sparsely observed random fields into a lower‑dimensional latent space and performs posterior sampling there. A variance‑preserving bridge controls local covariance inflation, while flow matching learns the latent prior. Experiments on Gaussian benchmarks, a bimodal posterior, and a Navier‑Stokes field show that LAPS accurately approximates posterior distributions, preserves multiple modes, and keeps conditional sampling costs nearly constant as spatial resolution increases.
arXiv:2606. 25882v1 Announce Type: new Abstract: DGPs are probabilistic models with remarkable prediction performance that concatenate GPs across several layers.
arXiv:2608. 15144v1 Announce Type: cross Abstract: Posterior sampling with a pretrained diffusion prior is governed by a conditional score whose intermediate likelihood component is generally intractable.
The paper introduces Diffusion Waltz, an MCMC algorithm that uses SDEdit-style noising and denoising as a proposal and corrects it with Metropolis-Hastings to achieve exact posterior sampling without evaluating the prior. It also incorporates observations into the proposal via a gradient‑free ensemble Kalman update while maintaining exactness. In experiments on a non‑differentiable Navier‑Stokes initial‑condition recovery problem, Diffusion Waltz outperforms existing baselines across various noise and nonlinearity regimes.
arXiv:2610.03503v1 Announce Type: new Abstract: Training-free posterior sampling methods, also known as Plug-and-Play methods, leverage pretrained unconditional diffusion or flow-matching models to s...
STITCH-OPE is a model‑based generative framework that uses denoising diffusion to perform off‑policy evaluation (OPE) in high‑dimensional, long‑horizon settings. It generates synthetic trajectories for a target policy by guiding a diffusion model trained on behavior data, subtracting the behavior policy’s score to avoid over‑regularization and stitching partial trajectories to extend horizon length. The authors provide theoretical variance‑reduction guarantees and demonstrate improved mean squared error, correlation, and regret on D4RL and OpenAI Gym benchmarks.
arXiv:2606. 26592v1 Announce Type: cross Abstract: We propose latent-space diffusion posterior sampling (L-DPS), an approximate Bayesian framework for high-dimensional inverse problems governed by partial differential equations (PDEs).
arXiv:2606. 03212v1 Announce Type: new Abstract: Low-rank tensor decomposition (TD) is usually effective on clean, fully observed data, but it often degrades under severe missingness or noise.
The paper introduces a data‑free learning objective called relative trajectory balance for training diffusion models to sample from a posterior defined by a diffusion prior and an arbitrary black‑box constraint or likelihood. It proves asymptotic correctness of this objective and demonstrates its use across vision, language, and multimodal tasks, including classifier guidance, language infilling, and text‑to‑image generation. Additionally, the method is applied to continuous control with a score‑based behavior prior, achieving state‑of‑the‑art results in offline reinforcement learning.
The paper introduces RED-KLwSGS, a new posterior sampling method for Bayesian imaging inverse problems that combines exact Gaussian updates for the data variable with underdamped (kinetic) Langevin diffusion for the auxiliary variable, driven by a one‑shot denoising score. It achieves the same per‑iteration cost as existing Langevin‑within‑SGS samplers while providing non‑asymptotic Wasserstein‑2 convergence guarantees for strongly log‑concave priors. A further variant, Joint-RED-KLwSGS, applies kinetic Langevin diffusion to both variables, and experiments using denoising diffusion probabilistic models on FFHQ and ImageNet demonstrate faster convergence and high‑quality image reconstruction.
FlowSGS introduces a flow-based posterior sampling method that combines Split Gibbs Sampling (SGS) with Langevin dynamics for the likelihood step and Stochastic Interpolants (SI) for the prior step. By integrating a pretrained flow model into the prior step via SI's reverse-time SDE and a novel timestep correction, FlowSGS reduces the number of network evaluations compared to plug‑and‑play diffusion samplers. Experiments demonstrate state‑of‑the‑art performance on various inverse problems, including the first flow‑based solution to a nonlinear inverse problem (Fourier phase retrieval).