n-MeRCI: A new Metric to Evaluate the Correlation Between Predictive Uncertainty and True Error

TL;DR

This paper introduces n-MeRCI, a new metric designed to evaluate how well predictive uncertainty correlates with true error in deep learning, especially for robotics applications like depth estimation.

Contribution

The paper proposes a novel metric, n-MeRCI, for assessing the quality of uncertainty estimates in regression tasks with deep neural networks.

Findings

n-MeRCI effectively evaluates uncertainty quality in regression.

The metric is validated on toy data and monocular depth estimation.

It highlights the importance of uncertainty assessment in robotics applications.

Abstract

As deep learning applications are becoming more and more pervasive in robotics, the question of evaluating the reliability of inferences becomes a central question in the robotics community. This domain, known as predictive uncertainty, has come under the scrutiny of research groups developing Bayesian approaches adapted to deep learning such as Monte Carlo Dropout. Unfortunately, for the time being, the real goal of predictive uncertainty has been swept under the rug. Indeed, these approaches are solely evaluated in terms of raw performance of the network prediction, while the quality of their estimated uncertainty is not assessed. Evaluating such uncertainty prediction quality is especially important in robotics, as actions shall depend on the confidence in perceived information. In this context, the main contribution of this article is to propose a novel metric that is adapted to the evaluation of relative uncertainty assessment and directly applicable to regression with deep neural networks. To experimentally validate this metric, we evaluate it on a toy dataset and then apply it to the task of monocular depth estimation.