Detecting Out-of-Distribution Inputs in Deep Neural Networks Using an Early-Layer Output

Abstract

Deep neural networks achieve superior performance in challenging tasks such as image classification. However, deep classifiers tend to incorrectly classify out-of-distribution (OOD) inputs, which are inputs that do not belong to the classifier training distribution. Several approaches have been proposed to detect OOD inputs, but the detection task is still an ongoing challenge. In this paper, we propose a new OOD detection approach that can be easily applied to an existing classifier and does not need to have access to OOD samples. The detector is a one-class classifier trained on the output of an early layer of the original classifier fed with its original training set. We apply our approach to several low- and high-dimensional datasets and compare it to the state-of-the-art detection approaches. Our approach achieves substantially better results over multiple metrics.

arXiv comment15 pages, 8 figures

References in corpus (14)

• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• Deep Residual Learning for Image Recognition
• A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks
• LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop
• A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks
• One-Class Classification: Taxonomy of Study and Review of Techniques
• Deep Bayesian Active Learning with Image Data
• Visualizing Deep Neural Network Decisions: Prediction Difference Analysis
• Learning Confidence for Out-of-Distribution Detection in Neural Networks
• TurkerGaze: Crowdsourcing Saliency with Webcam based Eye Tracking
• Improving Reconstruction Autoencoder Out-of-distribution Detection with Mahalanobis Distance
• Do Deep Generative Models Know What They Don't Know?
• Safer Classification by Synthesis
• Out-of-Distribution Detection using Multiple Semantic Label Representations

Cited by in corpus (8)

• Detecting Out-of-Distribution Examples with In-distribution Examples and Gram Matrices
• Calibrating Deep Neural Network Classifiers on Out-of-Distribution Datasets
• Detecting Unknown DGAs without Context Information
• Improved Robustness to Open Set Inputs via Tempered Mixup
• A Multi-Scale Framework for Out-of-Distribution Detection in Dermoscopic Images
• MOOD: Multi-level Out-of-distribution Detection
• DOODLER: Determining Out-Of-Distribution Likelihood from Encoder Reconstructions
• Toward a Realistic Benchmark for Out-of-Distribution Detection