DropAttack: A Random Dropped Weight Attack Adversarial Training for Natural Language Understanding-Reference-Cited by-同舟云学术

DropAttack: A Random Dropped Weight Attack Adversarial Training for Natural Language Understanding

Author:

Ni Shiwen¹^ORCID,Li Jiawen²^ORCID,Yang Min¹^ORCID,Kao Hung-Yu³^ORCID

Affiliation:

1. Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China

2. Maritime College, Guangdong Ocean University, Zhanjiang, China

3. Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan

Funder

Ministry of Science and Technology, Taiwan

SIAT-LHXX AI and Education Joint Lab

Publisher

Institute of Electrical and Electronics Engineers (IEEE)

Subject

Electrical and Electronic Engineering,Acoustics and Ultrasonics,Computer Science (miscellaneous),Computational Mathematics

Link

Reference42 articles.

1. Adversarial weight perturbation helps robust generalization;wu;Proc Adv Neural Inf Process Syst,0

2. Roberta: A robustly optimized bert pretraining approach;liu,2019

3. Adversarial training for free;shafahi;Proc 33rd Int Conf Neural Inf Process Syst,0

4. ALBERT: A. lite BERT for self-supervised learning of language representations;lan;Proc Int Conf Learn Representations,0

5. Adversarial machine learning at scale;kurakin;Proc Int Conf Learn Representations,0

Cited by 1 articles. 订阅此论文施引文献订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献