Nikolai L. Kühne, Astrid H. F. Kitchen, Marie S. Jensen, Mikkel S. L. Brøndt, Martin Gonzalez, Christophe Biscio, Zheng-Hua Tan
{"title":"通过扩散模型检测和防御对自动语音识别的恶意攻击","authors":"Nikolai L. Kühne, Astrid H. F. Kitchen, Marie S. Jensen, Mikkel S. L. Brøndt, Martin Gonzalez, Christophe Biscio, Zheng-Hua Tan","doi":"arxiv-2409.07936","DOIUrl":null,"url":null,"abstract":"Automatic speech recognition (ASR) systems are known to be vulnerable to\nadversarial attacks. This paper addresses detection and defence against\ntargeted white-box attacks on speech signals for ASR systems. While existing\nwork has utilised diffusion models (DMs) to purify adversarial examples,\nachieving state-of-the-art results in keyword spotting tasks, their\neffectiveness for more complex tasks such as sentence-level ASR remains\nunexplored. Additionally, the impact of the number of forward diffusion steps\non performance is not well understood. In this paper, we systematically\ninvestigate the use of DMs for defending against adversarial attacks on\nsentences and examine the effect of varying forward diffusion steps. Through\ncomprehensive experiments on the Mozilla Common Voice dataset, we demonstrate\nthat two forward diffusion steps can completely defend against adversarial\nattacks on sentences. Moreover, we introduce a novel, training-free approach\nfor detecting adversarial attacks by leveraging a pre-trained DM. Our\nexperimental results show that this method can detect adversarial attacks with\nhigh accuracy.","PeriodicalId":501284,"journal":{"name":"arXiv - EE - Audio and Speech Processing","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Detecting and Defending Against Adversarial Attacks on Automatic Speech Recognition via Diffusion Models\",\"authors\":\"Nikolai L. Kühne, Astrid H. F. Kitchen, Marie S. Jensen, Mikkel S. L. Brøndt, Martin Gonzalez, Christophe Biscio, Zheng-Hua Tan\",\"doi\":\"arxiv-2409.07936\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Automatic speech recognition (ASR) systems are known to be vulnerable to\\nadversarial attacks. This paper addresses detection and defence against\\ntargeted white-box attacks on speech signals for ASR systems. While existing\\nwork has utilised diffusion models (DMs) to purify adversarial examples,\\nachieving state-of-the-art results in keyword spotting tasks, their\\neffectiveness for more complex tasks such as sentence-level ASR remains\\nunexplored. Additionally, the impact of the number of forward diffusion steps\\non performance is not well understood. In this paper, we systematically\\ninvestigate the use of DMs for defending against adversarial attacks on\\nsentences and examine the effect of varying forward diffusion steps. Through\\ncomprehensive experiments on the Mozilla Common Voice dataset, we demonstrate\\nthat two forward diffusion steps can completely defend against adversarial\\nattacks on sentences. Moreover, we introduce a novel, training-free approach\\nfor detecting adversarial attacks by leveraging a pre-trained DM. Our\\nexperimental results show that this method can detect adversarial attacks with\\nhigh accuracy.\",\"PeriodicalId\":501284,\"journal\":{\"name\":\"arXiv - EE - Audio and Speech Processing\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-12\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - EE - Audio and Speech Processing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.07936\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - EE - Audio and Speech Processing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.07936","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
众所周知,自动语音识别(ASR)系统容易受到对抗性攻击。本文论述了自动语音识别(ASR)系统如何检测和防御针对语音信号的白盒攻击。虽然现有工作利用扩散模型(DM)来净化对抗性示例,在关键词识别任务中取得了最先进的结果,但对于句子级 ASR 等更复杂的任务,其有效性仍有待探索。此外,人们对前向扩散步数对性能的影响也不甚了解。在本文中,我们系统地研究了如何使用 DM 来抵御对句子的恶意攻击,并考察了改变前向扩散步数的效果。通过对 Mozilla Common Voice 数据集的全面实验,我们证明了两个前向扩散步骤可以完全抵御对句子的恶意攻击。此外,我们还引入了一种新颖的、无需训练的方法,利用预先训练的 DM 来检测对抗性攻击。实验结果表明,这种方法可以高精度地检测出对抗性攻击。
Detecting and Defending Against Adversarial Attacks on Automatic Speech Recognition via Diffusion Models
Automatic speech recognition (ASR) systems are known to be vulnerable to
adversarial attacks. This paper addresses detection and defence against
targeted white-box attacks on speech signals for ASR systems. While existing
work has utilised diffusion models (DMs) to purify adversarial examples,
achieving state-of-the-art results in keyword spotting tasks, their
effectiveness for more complex tasks such as sentence-level ASR remains
unexplored. Additionally, the impact of the number of forward diffusion steps
on performance is not well understood. In this paper, we systematically
investigate the use of DMs for defending against adversarial attacks on
sentences and examine the effect of varying forward diffusion steps. Through
comprehensive experiments on the Mozilla Common Voice dataset, we demonstrate
that two forward diffusion steps can completely defend against adversarial
attacks on sentences. Moreover, we introduce a novel, training-free approach
for detecting adversarial attacks by leveraging a pre-trained DM. Our
experimental results show that this method can detect adversarial attacks with
high accuracy.