The surrogate Gibbs-posterior of a corrected stochastic MALA: Towards uncertainty quantification for neural networks

Дата публикации: 17-08-2026 20:26:00

MALA is a popular gradient-based Markov chain Monte Carlo method to access the Gibbs-posterior distribution. Stochastic MALA (sMALA) scales to large data sets, but changes the target distribution from the Gibbs-posterior to a surrogate posterior which only exploits a reduced sample size. We introduce a corrected stochastic MALA (csMALA) with a simple correction term for which distance between the resulting surrogate posterior and the original Gibbs-posterior decreases in the full sample size while retaining scalability. In a nonparametric regression model, we prove a PAC-Bayes oracle inequality for the surrogate posterior. Uncertainties can be quantified by sampling from the surrogate posterior. Focusing on Bayesian neural networks, we analyze the diameter and coverage of credible balls for shallow neural networks and we show optimal contraction rates for deep neural networks. Our credibility result is independent of the correction and can also be applied to the standard Gibbs-posterior. A simulation study in a high-dimensional parameter space demonstrates that an estimator drawn from csMALA based on its surrogate Gibbs-posterior indeed exhibits these advantages in practice.

Схожие новости

#	Наименование новости	Тональность	Информативность	Дата публикации
1	Finite Neural Networks as Mixtures of Gaussian Processes: From Provable Error Bounds to Prior Selection	0	4.23	17-08-2026
2	Cheap Bootstrap for Fast Uncertainty Quantification of Stochastic Gradient Descent	0	6.38	17-08-2026
3	Mixing times of data-augmentation Gibbs samplers for high-dimensional probit regression	0	8.78	17-08-2026
4	Error Analysis for Deep ReLU Feedforward Density-Ratio Estimation with Bregman Divergence	0	8.78	17-08-2026
5	A Data-Augmented Contrastive Learning Approach to Nonparametric Density Estimation	0	8.4	17-08-2026
6	Extending Mean-Field Variational Inference via Entropic Regularization: Theory and Computation	0	8.7	17-08-2026
7	Exploring Novel Uncertainty Quantification through Forward Intensity Function Modeling	0	8.74	17-08-2026
8	Statistical Learning Theory for Neural Operators	0	10.21	17-08-2026
9	Generative Bayesian Inference with GANs	0	6.62	17-08-2026

Классификация: Наука. Схожих патентов: 0. Схожих новостей: 9. Тональность: 0. Информативность: 8.32. Источник: jmlr.org.