# Sampling-free Epistemic Uncertainty Estimation Using Approximated Variance Propagation

## TL;DR

This paper introduces a sampling-free method for estimating epistemic uncertainty in neural networks, significantly reducing computational costs while maintaining high-quality uncertainty estimates for large-scale visual tasks.

## Contribution

It proposes an approximation technique for epistemic uncertainty that eliminates the need for sampling, improving efficiency over existing methods like Monte-Carlo dropout.

## Key findings

- Reduces computational overhead compared to sampling-based methods
- Maintains high-quality uncertainty estimates in large-scale visual tasks
- Demonstrates effectiveness in semantic segmentation and depth regression

## Abstract

We present a sampling-free approach for computing the epistemic uncertainty of a neural network. Epistemic uncertainty is an important quantity for the deployment of deep neural networks in safety-critical applications, since it represents how much one can trust predictions on new data. Recently promising works were proposed using noise injection combined with Monte-Carlo sampling at inference time to estimate this quantity (e.g. Monte-Carlo dropout). Our main contribution is an approximation of the epistemic uncertainty estimated by these methods that does not require sampling, thus notably reducing the computational overhead. We apply our approach to large-scale visual tasks (i.e., semantic segmentation and depth regression) to demonstrate the advantages of our method compared to sampling-based approaches in terms of quality of the uncertainty estimates as well as of computational overhead.