High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach
Machine Learning
2019-04-10 v3
Abstract
This paper considers the generation of prediction intervals (PIs) by neural networks for quantifying uncertainty in regression tasks. It is axiomatic that high-quality PIs should be as narrow as possible, whilst capturing a specified portion of data. We derive a loss function directly from this axiom that requires no distributional assumption. We show how its form derives from a likelihood principle, that it can be used with gradient descent, and that model uncertainty is accounted for in ensembled form. Benchmark experiments show the method outperforms current state-of-the-art uncertainty quantification methods, reducing average PI width by over 10%.
Cite
@article{arxiv.1802.07167,
title = {High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach},
author = {Tim Pearce and Mohamed Zaki and Alexandra Brintrup and Andy Neely},
journal= {arXiv preprint arXiv:1802.07167},
year = {2019}
}