EUVIMEDEUVIMED entdecken
Crossref

Low-Dimensional Gradient Helps Out-of-Distribution Detection

Yingwen Wu, Tao Li, Xinwen Cheng, Jie Yang, Xiaolin Huang

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2024-12 · 46 (12) · 11378-11391

DOI: 10.1109/tpami.2024.3459988 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024-12. Verfasst von Yingwen Wu und 4 weiteren Autor:innen. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 69 Literaturangaben. Gefördert von: National Key Research Development Project, National Natural Science Foundation of China, Shanghai Municipal Science and Technology Research Program.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Detecting out-of-distribution (OOD) samples is essential for ensuring the reliability of deep neural networks (DNNs) in real-world scenarios. While previous research has predominantly investigated the disparity between in-distribution (ID) and OOD data through forward information analysis, the discrepancy in parameter gradients during the backward process of DNNs has received insufficient attention. Existing studies on gradient disparities mainly focus on the utilization of gradient norms, neglecting the wealth of information embedded in gradient directions. To bridge this gap, in this paper, we conduct a comprehensive investigation into leveraging the entirety of gradient information for OOD detection. The primary challenge arises from the high dimensionality of gradients due to the large number of network parameters. To solve this problem, we propose performing linear dimension reduction on the gradient using a designated subspace that comprises principal components. This innovative technique enables us to obtain a low-dimensional representation of the gradient with minimal information loss. Subsequently, by integrating the reduced gradient with various existing detection score functions, our approach demonstrates superior performance across a wide range of detection tasks. For instance, on the ImageNet benchmark with ResNet50 model, our method achieves an average reduction of 11.15 % in the false positive rate at 95 % recall (FPR95) compared to the current state-of-the-art approach.

Förderung

  • National Key Research Development Project
  • National Natural Science Foundation of China
  • Shanghai Municipal Science and Technology Research Program
  • Major Project

Zitieren

APA

Wu, Y., Li, T., Cheng, X., Yang, J., & Huang, X. (2024). Low-Dimensional Gradient Helps Out-of-Distribution Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12), 11378-11391. https://doi.org/10.1109/tpami.2024.3459988

Vancouver

Wu Y, Li T, Cheng X, Yang J, Huang X. Low-Dimensional Gradient Helps Out-of-Distribution Detection. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2024;46(12):11378-11391. doi:10.1109/tpami.2024.3459988

BibTeX
@article{Wu2024,
  author = {Wu, Yingwen and Li, Tao and Cheng, Xinwen and Yang, Jie and Huang, Xiaolin},
  title = {Low-Dimensional Gradient Helps Out-of-Distribution Detection},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2024},
  volume = {46},
  number = {12},
  pages = {11378--11391},
  doi = {10.1109/tpami.2024.3459988}
}
Kostenlos starten →