Uncertainty separation via ensemble quantile regression

TL;DR

This paper presents a scalable ensemble quantile regression framework that improves uncertainty estimation and separation in data-driven modeling, outperforming existing methods like Deep Ensembles and MC dropout.

Contribution

It introduces a novel ensemble quantile regression approach with an iterative algorithm for better separation of aleatoric and epistemic uncertainties, scalable to large datasets.

Findings

Superior uncertainty estimation over competing methods

Effective separation of uncertainty types in high-uncertainty regions

Robust performance demonstrated on synthetic benchmarks

Abstract

This paper introduces a novel and scalable framework for uncertainty estimation and separation with applications in data driven modeling in science and engineering tasks where reliable uncertainty quantification is critical. Leveraging an ensemble of quantile regression (E-QR) models, our approach enhances aleatoric uncertainty estimation while preserving the quality of epistemic uncertainty, surpassing competing methods, such as Deep Ensembles (DE) and Monte Carlo (MC) dropout. To address challenges in separating uncertainty types, we propose an algorithm that iteratively improves separation through progressive sampling in regions of high uncertainty. Our framework is scalable to large datasets and demonstrates superior performance on synthetic benchmarks, offering a robust tool for uncertainty quantification in data-driven applications.