Research Area:  Machine Learning

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.

Keywords:  
Deep Deterministic Neural Network
Neural Network
Machine Learning
Deep Ensembles
RBF networks

Author(s) Name:  Joost Van Amersfoort, Lewis Smith, Yee Whye Teh, Yarin Gal

Publisher name:  PMLR

DOI:  

Volume Information:   Volume 119

Paper Link:   https://proceedings.mlr.press/v119/van-amersfoort20a.html