Generative Modeling with Bayesian Sample Inference

2025/02/11 by Marten Lienen, Lienen, Marten, Marcel Kollovieh +3 · 2 citations
Computer Science · #Bayesian Methods and Mixture Models #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · pdf · doi:10.48550/arxiv.2502.07580

openalex publication_date 2025/02/11 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We derive a novel generative model from iterative Gaussian posterior inference. By treating the generated sample as an unknown variable, we can formulate the sampling process in the language of Bayesian probability. Our model uses a sequence of prediction and posterior update steps to iteratively narrow down the unknown sample starting from a broad initial belief. In addition to a rigorous theoretical analysis, we establish a connection between our model and diffusion models and show that it includes Bayesian Flow Networks (BFNs) as a special case. In our experiments, we demonstrate that our model improves sample quality on ImageNet32 over both BFNs and the closely related Variational Diffusion Models, while achieving equivalent log-likelihoods on ImageNet32 and ImageNet64. Find our code at https://github.com/martenlienen/bsi.

Cited by

• Discrete Bayesian Sample Inference for Graph Generation
• Edit-Based Flow Matching for Temporal Point Processes

Related

• Bayesian Flow Networks
• Posterior Mean Matching: Generative Modeling through Online Bayesian Inference
• Generative Uncertainty in Diffusion Models
• Unifying Bayesian Flow Networks and Diffusion Models through Stochastic Differential Equations
• Outsourced diffusion sampling: Efficient posterior inference in latent spaces of generative models