Logit Perturbation
Paper
Paper/Presentation Title | Logit Perturbation |
---|---|
Presentation Type | Paper |
Authors | Li, Mengyang, Su, Fengguang, Wu, Ou and Zhang, Ji |
Journal or Proceedings Title | Proceedings of the 36th AAAI Conference on Artificial Intelligence (AAAI 2022) |
Journal Citation | 36, pp. 1359-1366 |
Number of Pages | 8 |
Year | 2022 |
Place of Publication | United Sates |
ISBN | 1577358767 |
9781-577358763 | |
Digital Object Identifier (DOI) | https://doi.org/10.1609/aaai.v36i2.20024 |
Web Address (URL) of Paper | https://ojs.aaai.org/index.php/AAAI/article/view/20024 |
Web Address (URL) of Conference Proceedings | https://ojs.aaai.org/index.php/AAAI/issue/view/508 |
Conference/Event | 36th AAAI Conference on Artificial Intelligence (AAAI 2022) |
Event Details | 36th AAAI Conference on Artificial Intelligence (AAAI 2022) Parent AAAI Conference on Artificial Intelligence Delivery Online Event Date 22 Feb 2022 to end of 01 Mar 2022 |
Abstract | Features, logits, and labels are the three primary data when a sample passes through a deep neural network. Feature perturbation and label perturbation receive increasing attention in recent years. They have been proven to be useful in various deep learning approaches. For example, (adversarial) feature perturbation can improve the robustness or even generalization capability of learned models. However, limited studies have explicitly explored for the perturbation of logit vectors. This work discusses several existing methods related to logit perturbation. Based on a unified viewpoint between positive/negative data augmentation and loss variations incurred by logit perturbation, a new method is proposed to explicitly learn to perturb logits. A comparative analysis is conducted for the perturbations used in our and existing methods. Extensive experiments on benchmark image classification data sets and their long-tail versions indicated the competitive performance of our learning method. In addition, existing methods can be further improved by utilizing our method. |
Keywords | Computer Vision (CV) |
ANZSRC Field of Research 2020 | 460299. Artificial intelligence not elsewhere classified |
Public Notes | Files associated with this item cannot be displayed due to copyright restrictions. |
Byline Affiliations | Tianjin University, China |
University of Southern Queensland |
https://research.usq.edu.au/item/z58z7/logit-perturbation
53
total views2
total downloads4
views this month0
downloads this month