AI RESEARCH
Large-scale Score-based Variational Posterior Inference for Bayesian Deep Neural Networks
arXiv CS.LG
•
ArXi:2602.05873v2 Announce Type: replace Bayesian (deep) neural networks (BNN) are often attractive than the vanilla point-estimate deep learning in various aspects including uncertainty quantification, robustness to noise, resistance to overfitting, and more. The variational inference (VI) is one of the most widely adopted approximate inference methods. Whereas the ELBO-based variational free energy method is a dominant choice in the literature, in this paper we