Uncertainty Estimation Using a Single Deep Deterministic Neural Network

TL;DR

This paper introduces DUQ, a deterministic neural network method that efficiently detects out-of-distribution data with a single forward pass, matching or surpassing ensemble methods in accuracy.

Contribution

The paper presents DUQ, a novel approach combining RBF network ideas with a new loss and gradient penalty, enabling reliable out-of-distribution detection in a single model.

Findings

Matches softmax model accuracy

Outperforms or matches Deep Ensembles in OOD detection

Scales well to large datasets

Abstract

We propose a method for training a deterministic deep model that can find and reject out of distribution data points at test time with a single forward pass. Our approach, deterministic uncertainty quantification (DUQ), builds upon ideas of RBF networks. We scale training in these with a novel loss function and centroid updating scheme and match the accuracy of softmax models. By enforcing detectability of changes in the input using a gradient penalty, we are able to reliably detect out of distribution data. Our uncertainty quantification scales well to large datasets, and using a single model, we improve upon or match Deep Ensembles in out of distribution detection on notable difficult dataset pairs such as FashionMNIST vs. MNIST, and CIFAR-10 vs. SVHN.