中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 自动化研究所
  3. 中国科学院自动化研究所
  4. 多模态人工智能系统全国重点实验室
Disentangled Text Representation Learning With Information-Theoretic Perspective for Adversarial Robustness

文献类型：期刊论文

作者Zhao, Jiahao1,2; Mao, Wenji1,2; Zeng, Daniel Dajun1,2
刊名IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING
出版日期2024
卷号32页码:1237-1247
关键词Adversarial robustness variation of information disentangled text representation learning
ISSN号2329-9290
DOI10.1109/TASLP.2024.3358052
通讯作者Mao, Wenji(wenji.mao@ia.ac.cn)
英文摘要Adversarial vulnerability remains a major obstacle to the construction of reliable NLP systems. When imperceptible perturbations are added to raw input text, the performance of a deep learning model may drop dramatically under attacks. Recent work has argued that the adversarial vulnerability of a model is caused by non-robust features in supervised training. Thus, in this paper, we tackle the adversarial robustness challenge by means of disentangled representation learning, which is able to explicitly disentangle robust and non-robust features in text. Specifically, inspired by the variation of information (VI) in information theory, we derive a disentangled learning objective composed of mutual information to represent both the semantic representativeness of latent embeddings and the differentiation of robust and non-robust features. On the basis of this, we design a disentangled learning network to estimate the mutual information for realization. Experiments on the typical text-based tasks show that our method significantly outperforms the representative methods under adversarial attacks, indicating that discarding non-robust features is critical for improving model robustness.
资助项目Ministry of Science and Technology of China
WOS研究方向Acoustics ; Engineering
语种英语
WOS记录号WOS:001174088200002
出版者IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
资助机构Ministry of Science and Technology of China
源URL[http://ir.ia.ac.cn/handle/173211/57888]  
专题多模态人工智能系统全国重点实验室
通讯作者Mao, Wenji
作者单位1.Chinese Acad Sci, Inst Automat, State Key Lab Multimodal Artificial Intelligence S, Beijing 100190, Peoples R China
2.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China
推荐引用方式
GB/T 7714		 Zhao, Jiahao,Mao, Wenji,Zeng, Daniel Dajun. Disentangled Text Representation Learning With Information-Theoretic Perspective for Adversarial Robustness[J]. IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING,2024,32:1237-1247.
APA Zhao, Jiahao,Mao, Wenji,&Zeng, Daniel Dajun.(2024).Disentangled Text Representation Learning With Information-Theoretic Perspective for Adversarial Robustness.IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING,32,1237-1247.
MLA Zhao, Jiahao,et al."Disentangled Text Representation Learning With Information-Theoretic Perspective for Adversarial Robustness".IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING 32(2024):1237-1247.

入库方式： OAI收割

来源：自动化研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。