A New Ensemble Adversarial Attack Powered by Long-term Gradient Memories
Abstract
Deep neural networks are vulnerable to adversarial attacks.
- Publication:
-
arXiv e-prints
- Pub Date:
- November 2019
- DOI:
- arXiv:
- arXiv:1911.07682
- Bibcode:
- 2019arXiv191107682C
- Keywords:
-
- Computer Science - Machine Learning;
- Statistics - Machine Learning
- E-Print:
- Accepted by AAAI2020