On Feynman--Kac training of partial Bayesian neural networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Zheng, Mair, Sebastian, Schön, Thomas B., Sjölund, Jens
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909120739672064
author Zhao, Zheng
Mair, Sebastian
Schön, Thomas B.
Sjölund, Jens
author_facet Zhao, Zheng
Mair, Sebastian
Schön, Thomas B.
Sjölund, Jens
contents Recently, partial Bayesian neural networks (pBNNs), which only consider a subset of the parameters to be stochastic, were shown to perform competitively with full Bayesian neural networks. However, pBNNs are often multi-modal in the latent variable space and thus challenging to approximate with parametric models. To address this problem, we propose an efficient sampling-based training strategy, wherein the training of a pBNN is formulated as simulating a Feynman--Kac model. We then describe variations of sequential Monte Carlo samplers that allow us to simultaneously estimate the parameters and the latent posterior distribution of this model at a tractable computational cost. Using various synthetic and real-world datasets we show that our proposed training scheme outperforms the state of the art in terms of predictive performance.
format Preprint
id arxiv_https___arxiv_org_abs_2310_19608
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle On Feynman--Kac training of partial Bayesian neural networks
Zhao, Zheng
Mair, Sebastian
Schön, Thomas B.
Sjölund, Jens
Machine Learning
Recently, partial Bayesian neural networks (pBNNs), which only consider a subset of the parameters to be stochastic, were shown to perform competitively with full Bayesian neural networks. However, pBNNs are often multi-modal in the latent variable space and thus challenging to approximate with parametric models. To address this problem, we propose an efficient sampling-based training strategy, wherein the training of a pBNN is formulated as simulating a Feynman--Kac model. We then describe variations of sequential Monte Carlo samplers that allow us to simultaneously estimate the parameters and the latent posterior distribution of this model at a tractable computational cost. Using various synthetic and real-world datasets we show that our proposed training scheme outperforms the state of the art in terms of predictive performance.
title On Feynman--Kac training of partial Bayesian neural networks
topic Machine Learning
url https://arxiv.org/abs/2310.19608