Latent Derivative Bayesian Last Layer Networks

Loading...
Thumbnail Image

Access rights

openAccess
publishedVersion

URL

Journal Title

Journal ISSN

Volume Title

A4 Artikkeli konferenssijulkaisussa

Date

2021

Major/Subject

Mcode

Degree programme

Language

en

Pages

13

Series

24TH INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS (AISTATS), pp. 1198-1206, Proceedings of Machine Learning Research ; Volume 130

Abstract

Bayesian neural networks (BNN) are powerful parametric models for nonlinear regression with uncertainty quantification. However, the approximate inference techniques for weight space priors suffer from several drawbacks. The 'Bayesian last layer' (BLL) is an alternative BNN approach that learns the feature space for an exact Bayesian linear model with explicit predictive distributions. However, its predictions outside of the data distribution (OOD) are typically overconfident, as the marginal likelihood objective results in a learned feature space that overfits to the data. We overcome this weakness by introducing a functional prior on the model's derivatives w.r.t. the inputs. Treating these Jacobians as latent variables, we incorporate the prior into the objective to influence the smoothness and diversity of the features, which enables greater predictive uncertainty. For the BLL, the Jacobians can be computed directly using forward mode automatic differentiation, and the distribution over Jacobians may be obtained in closed-form. We demonstrate this method enhances the BLL to Gaussian process-like performance on tasks where calibrated uncertainty is critical: OOD regression, Bayesian optimization and active learning, which include high-dimensional real-world datasets.

Description

Keywords

Other note

Citation

Watson, J, Lin, J A, Klink, P, Pajarinen, J & Peters, J 2021, Latent Derivative Bayesian Last Layer Networks . in A Banerjee & K Fukumizu (eds), 24TH INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE AND STATISTICS (AISTATS) . Proceedings of Machine Learning Research, vol. 130, JMLR, pp. 1198-1206, International Conference on Artificial Intelligence and Statistics, Virtual, Online, 13/04/2021 . < http://proceedings.mlr.press/v130/watson21a/watson21a.pdf >