RG: OUT-OF-DISTRIBUTION DETECTION WITH REACTIVATE GRADNORM
Lists
Only you see this
Abstract
Detecting out-of-distribution (OOD) data is critical to building reliable machine learning systems in the open world. Previous works mainly perform OOD detection in feature space or output space. Recently, researchers have achieved promising results using gradient information, which combines the information in both feature and output space for OOD detection. However, existing works still suffer from the problem of overconfidence. To address this problem, we propose a novel method called ``Reactivate Gradnorm (RG)'', which exploits the norm of the clipped feature vector and the energy in the output space for OOD detection. To verify the effectiveness of our method, we conduct experiments on four benchmark datasets. Experimental results demonstrate that our RG outperforms existing state-of-the-art approaches by 2.06\% in average AUROC. Meanwhile, RG is easy to implement and does not require additional OOD data or fine-tuning process. We can realize OOD detection in only one forward pass of any pretrained model.
BibTeX
@inproceedings{xu2023rg,
author = {Mingyu Xu and Kexin Wang and Zheng Lian and Licai Sun and Bin Liu and Jianhua Tao},
title = {{RG: OUT-OF-DISTRIBUTION DETECTION WITH REACTIVATE GRADNORM}},
booktitle = {International Conference on Learning Representations},
year = {2023},
url = {https://openreview.net/forum?id=-hMNEMgT8Wd},
}Loading related content...