Variational Bayesian Last Layers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Harrison, James, Willes, John, Snoek, Jasper
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909172906328064
author Harrison, James
Willes, John
Snoek, Jasper
author_facet Harrison, James
Willes, John
Snoek, Jasper
contents We introduce a deterministic variational formulation for training Bayesian last layer neural networks. This yields a sampling-free, single-pass model and loss that effectively improves uncertainty estimation. Our variational Bayesian last layer (VBLL) can be trained and evaluated with only quadratic complexity in last layer width, and is thus (nearly) computationally free to add to standard architectures. We experimentally investigate VBLLs, and show that they improve predictive accuracy, calibration, and out of distribution detection over baselines across both regression and classification. Finally, we investigate combining VBLL layers with variational Bayesian feature learning, yielding a lower variance collapsed variational inference method for Bayesian neural networks.
format Preprint
id arxiv_https___arxiv_org_abs_2404_11599
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Variational Bayesian Last Layers
Harrison, James
Willes, John
Snoek, Jasper
Machine Learning
Computer Vision and Pattern Recognition
We introduce a deterministic variational formulation for training Bayesian last layer neural networks. This yields a sampling-free, single-pass model and loss that effectively improves uncertainty estimation. Our variational Bayesian last layer (VBLL) can be trained and evaluated with only quadratic complexity in last layer width, and is thus (nearly) computationally free to add to standard architectures. We experimentally investigate VBLLs, and show that they improve predictive accuracy, calibration, and out of distribution detection over baselines across both regression and classification. Finally, we investigate combining VBLL layers with variational Bayesian feature learning, yielding a lower variance collapsed variational inference method for Bayesian neural networks.
title Variational Bayesian Last Layers
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2404.11599