SOTAVerified

Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty

2019-06-28NeurIPS 2019Code Available0· sign in to hype

Dan Hendrycks, Mantas Mazeika, Saurav Kadavath, Dawn Song

arXivPDF

Code Available — Be the first to reproduce this paper.

Reproduce

Code

Abstract

Self-supervision provides effective representations for downstream tasks without requiring labels. However, existing approaches lag behind fully supervised training and are often not thought beneficial beyond obviating or reducing the need for annotations. We find that self-supervision can benefit robustness in a variety of ways, including robustness to adversarial examples, label corruption, and common input corruptions. Additionally, self-supervision greatly benefits out-of-distribution detection on difficult, near-distribution outliers, so much so that it exceeds the performance of fully supervised methods. These results demonstrate the promise of self-supervision for improving robustness and uncertainty estimation and establish these tasks as new axes of evaluation for future self-supervised learning research.

Tasks

Anomaly DetectionOutlier DetectionOut-of-Distribution DetectionSelf-Supervised LearningUnsupervised Anomaly Detection

Benchmark Results

DatasetModelMetricClaimedVerifiedStatus
Anomaly Detection on Anomaly Detection on Unlabeled ImageNet-30 vs Flowers-102ROT+TransROC-AUC86.3—Unverified
Anomaly Detection on Unlabeled ImageNet-30 vs CUB-200ROT+TransROC-AUC74.5—Unverified
One-class CIFAR-10SSOODAUROC90.1—Unverified
One-class ImageNet-30RotNet + Self-AttentionAUROC81.6—Unverified
One-class ImageNet-30RotNet + TranslationAUROC77.9—Unverified
One-class ImageNet-30Supervised (OE)AUROC56.1—Unverified
One-class ImageNet-30RotNetAUROC65.3—Unverified
One-class ImageNet-30RotNet + Translation + Self-Attention + ResizeAUROC85.7—Unverified
One-class ImageNet-30RotNet + Translation + Self-AttentionAUROC84.8—Unverified

Reproductions