Improved uncertainty quantification for neural networks with Bayesian last layer

TL;DR

This paper introduces an efficient reformulation for training Bayesian last layer neural networks, improving uncertainty quantification especially for extrapolation points, and demonstrates superior predictive performance over existing Bayesian methods.

Contribution

It presents a novel reformulation of the log-marginal likelihood enabling efficient training of Bayesian last layer neural networks with improved uncertainty quantification.

Findings

Achieves highest log-predictive density compared to Bayesian linear regression and variational Bayesian neural networks.

Provides a metric and method to better quantify uncertainty for extrapolation points.

Demonstrates effectiveness in a simulation study with multivariate data.

Abstract

Uncertainty quantification is an important task in machine learning - a task in which standardneural networks (NNs) have traditionally not excelled. This can be a limitation for safety-critical applications, where uncertainty-aware methods like Gaussian processes or Bayesian linear regression are often preferred. Bayesian neural networks are an approach to address this limitation. They assume probability distributions for all parameters and yield distributed predictions. However, training and inference are typically intractable and approximations must be employed. A promising approximation is NNs with Bayesian last layer (BLL). They assume distributed weights only in the linear output layer and yield a normally distributed prediction. To approximate the intractable Bayesian neural network, point estimates of the distributed weights in all but the last layer should be obtained by maximizing the marginal likelihood. This has previously been challenging, as the marginal likelihood is expensive to evaluate in this setting. We present a reformulation of the log-marginal likelihood of a NN with BLL which allows for efficient training using backpropagation. Furthermore, we address the challenge of uncertainty quantification for extrapolation points. We provide a metric to quantify the degree of extrapolation and derive a method to improve the uncertainty quantification for these points. Our methods are derived for the multivariate case and demonstrated in a simulation study. In comparison to Bayesian linear regression with fixed features, and a Bayesian neural network trained with variational inference, our proposed method achieves the highest log-predictive density on test data.