Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks

Shiyu Liang, Yixuan Li, R. Srikant
6/8/2017
cs.LGstat.ML

Official code: facebookresearch/odin (545 stars, last commit 2021-10-12).

Abstract

We consider the problem of detecting out-of-distribution images in neural networks. We propose ODIN, a simple and effective method that does not require any change to a pre-trained neural network. Our method is based on the observation that using temperature scaling and adding small perturbations to the input can separate the softmax score distributions between in- and out-of-distribution images, allowing for more effective detection. We show in a series of experiments that ODIN is compatible with diverse network architectures and datasets. It consistently outperforms the baseline approach by a large margin, establishing a new state-of-the-art performance on this task. For example, ODIN reduces the false positive rate from the baseline 34.7% to 4.3% on the DenseNet (applied to CIFAR-10) when the true positive rate is 95%.

View on arXivView PDF

Code Implementations(1)

facebookresearch/odinOfficial100%

A simple and effective method for detecting out-of-distribution images in neural networks.

545103PythonFeb 10, 20185 years agoNOASSERTION

Similar Papers

WeiPer: OOD Detection using Weight Perturbations of Class Projections

89% match1 repo22024

A Simple Test-Time Method for Out-of-Distribution Detection

89% match2022

Energy-based Out-of-distribution Detection

88% match1 repo4412020

Out-of-Distribution Detection using Neural Activation Prior

88% match2024

A Functional Data Perspective and Baseline On Multi-Layer Out-of-Distribution Detection

87% match2023

Learning Confidence for Out-of-Distribution Detection in Neural Networks

87% match1 repo2732018

Cite this paper

@article{liang2017enhancing,
  title  = {Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks},
  author = {Shiyu Liang and Yixuan Li and R. Srikant},
  year   = {2017},
  eprint = {1706.02690},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1706.02690}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1706.02690.svg)](https://codeofpaper.com/paper/1706.02690)

Discussion