INOR-An Intelligent noise reduction method to defend against adversarial audio examples

doi:10.1016/j.neucom.2020.02.110

	INOR-An Intelligent noise reduction method to defend against adversarial audio examples
	Guo, Qingli 1,2; Ye, Jing 1,2; Chen, Yiran 3; Hu, Yu 1,2; Lan, Yazhu 1; Zhang, Guohe 4; Li, Xiaowei 1,2
	2020-08-11
发表期刊	NEUROCOMPUTING
ISSN	0925-2312
卷号	401 页码:160-172
摘要	Recently, Automatic Speech Recognition(ASR) systems are seriously threatened by adversarial audio examples. The defense against adversarial audio examples has become an urgent issue. Different from adversarial image examples whose target is limited in the finite categories, the target of adversarial audio examples can be any combination of the words in a language. Adversarial audio examples aim to change the semantic of the audio. The semantic is explicitly represented in transcription distance, which affects the adversarial perturbation. This paper analyzes the relationship between semantic difference and adversarial perturbation. Quantization and local smoothing are calibrated to evaluate their performance. We observe that, for adversarial audio examples with different transcription distance levels, the capability of different denoising strategies varies. Therefore, we first introduce the wavelet filter, which denoises the signal in the transformed domain. Then we explore the defense capability of combined filters. Finally, a new intelligent noise reduction method-INOR is proposed to improve the denoising performance of audios under different levels of transcription distance. Experimental results show that INOR is effective in mitigating the adversarial perturbations for adversarial examples with different transcription distance levels. The average CER and WER is reduced by 33% and 55%. (C) 2020 Elsevier B.V. All rights reserved.
关键词	Adversarial audio examples Defense against adversarial audio examples INOR
DOI	10.1016/j.neucom.2020.02.110
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China (NSFC)[61532017] ; National Natural Science Foundation of China (NSFC)[61704174] ; National Natural Science Foundation of China (NSFC)[61432017] ; National Natural Science Foundation of China (NSFC)[61521092]
WOS研究方向	Computer Science
WOS类目	Computer Science, Artificial Intelligence
WOS记录号	WOS:000544725700016
出版者	ELSEVIER
引用统计	被引频次：5[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/15067
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Guo, Qingli
作者单位	1.Chinese Acad Sci, Inst Comp Technol, State Key Lab Comp Architecture, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China 3.Duke Univ, Dept Elect & Comp Engn, Durham, NC 27708 USA 4.Xi An Jiao Tong Univ, Sch Microelect, Xian 710049, Shanxi, Peoples R China
推荐引用方式 GB/T 7714	Guo, Qingli,Ye, Jing,Chen, Yiran,et al. INOR-An Intelligent noise reduction method to defend against adversarial audio examples[J]. NEUROCOMPUTING,2020,401:160-172.
APA	Guo, Qingli.,Ye, Jing.,Chen, Yiran.,Hu, Yu.,Lan, Yazhu.,...&Li, Xiaowei.(2020).INOR-An Intelligent noise reduction method to defend against adversarial audio examples.NEUROCOMPUTING,401,160-172.
MLA	Guo, Qingli,et al."INOR-An Intelligent noise reduction method to defend against adversarial audio examples".NEUROCOMPUTING 401(2020):160-172.