Detecting Out-of-Distribution Examples with In-distribution Examples and Gram Matrices
arXiv:1912.12510
Abstract
When presented with Out-of-Distribution (OOD) examples, deep neural networks yield confident, incorrect predictions. Detecting OOD examples is challenging, and the potential risks are high. In this paper, we propose to detect OOD examples by identifying inconsistencies between activity patterns and class predicted. We find that characterizing activity patterns by Gram matrices and identifying anomalies in gram matrix values can yield high OOD detection rates. We identify anomalies in the gram matrices by simply comparing each value with its respective range observed over the training data. Unlike many approaches, this can be used with any pre-trained softmax classifier and does not require access to OOD data for fine-tuning hyperparameters, nor does it require OOD access for inferring parameters. The method is applicable across a variety of architectures and vision datasets and, for the important and surprisingly hard task of detecting far-from-distribution out-of-distribution examples, it generally performs better than or equal to state-of-the-art OOD detection methods (including those that do assume access to OOD examples).
NeurIPS 2019 Workshop on Safety and Robustness in Decision Making
References in corpus (6)
- Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
- Deep Anomaly Detection with Outlier Exposure
- Out-of-distribution Detection in Classifiers via Generation
- Detecting Out-of-Distribution Inputs in Deep Neural Networks Using an Early-Layer Output
- Unsupervised Out-of-Distribution Detection with Batch Normalization
- Out-of-Distribution Detection Using Neural Rendering Generative Models