Characterizing audio adversarial examples using temporal dependency

Zhuolin Yang, Pin Yu Chen, Bo Li, Dawn Song

Research output: Contribution to conferencePaper

Abstract

Recent studies have highlighted adversarial examples as a ubiquitous threat to different neural network models and many downstream applications. Nonetheless, as unique data properties have inspired distinct and powerful learning principles, this paper aims to explore their potentials towards mitigating adversarial inputs. In particular, our results reveal the importance of using the temporal dependency in audio data to gain discriminate power against adversarial examples. Tested on the automatic speech recognition (ASR) tasks and three recent audio adversarial attacks, we find that (i) input transformation developed from image adversarial defense provides limited robustness improvement and is subtle to advanced attacks; (ii) temporal dependency can be exploited to gain discriminative power against audio adversarial examples and is resistant to adaptive attacks considered in our experiments. Our results not only show promising means of improving the robustness of ASR systems but also offer novel insights in exploiting domain-specific data properties to mitigate negative effects of adversarial examples.

Original languageEnglish (US)
StatePublished - Jan 1 2019
Event7th International Conference on Learning Representations, ICLR 2019 - New Orleans, United States
Duration: May 6 2019May 9 2019

Conference

Conference7th International Conference on Learning Representations, ICLR 2019
CountryUnited States
CityNew Orleans
Period5/6/195/9/19

Fingerprint

Speech recognition
Neural networks
neural network
Experiments
threat
Attack
experiment
learning
Automatic Speech Recognition
Robustness

ASJC Scopus subject areas

  • Education
  • Computer Science Applications
  • Linguistics and Language
  • Language and Linguistics

Cite this

Yang, Z., Chen, P. Y., Li, B., & Song, D. (2019). Characterizing audio adversarial examples using temporal dependency. Paper presented at 7th International Conference on Learning Representations, ICLR 2019, New Orleans, United States.

Characterizing audio adversarial examples using temporal dependency. / Yang, Zhuolin; Chen, Pin Yu; Li, Bo; Song, Dawn.

2019. Paper presented at 7th International Conference on Learning Representations, ICLR 2019, New Orleans, United States.

Research output: Contribution to conferencePaper

Yang, Z, Chen, PY, Li, B & Song, D 2019, 'Characterizing audio adversarial examples using temporal dependency' Paper presented at 7th International Conference on Learning Representations, ICLR 2019, New Orleans, United States, 5/6/19 - 5/9/19, .
Yang Z, Chen PY, Li B, Song D. Characterizing audio adversarial examples using temporal dependency. 2019. Paper presented at 7th International Conference on Learning Representations, ICLR 2019, New Orleans, United States.
Yang, Zhuolin ; Chen, Pin Yu ; Li, Bo ; Song, Dawn. / Characterizing audio adversarial examples using temporal dependency. Paper presented at 7th International Conference on Learning Representations, ICLR 2019, New Orleans, United States.
@conference{f0c039cb71b24b4a9fdc07f8f8d57b52,
title = "Characterizing audio adversarial examples using temporal dependency",
abstract = "Recent studies have highlighted adversarial examples as a ubiquitous threat to different neural network models and many downstream applications. Nonetheless, as unique data properties have inspired distinct and powerful learning principles, this paper aims to explore their potentials towards mitigating adversarial inputs. In particular, our results reveal the importance of using the temporal dependency in audio data to gain discriminate power against adversarial examples. Tested on the automatic speech recognition (ASR) tasks and three recent audio adversarial attacks, we find that (i) input transformation developed from image adversarial defense provides limited robustness improvement and is subtle to advanced attacks; (ii) temporal dependency can be exploited to gain discriminative power against audio adversarial examples and is resistant to adaptive attacks considered in our experiments. Our results not only show promising means of improving the robustness of ASR systems but also offer novel insights in exploiting domain-specific data properties to mitigate negative effects of adversarial examples.",
author = "Zhuolin Yang and Chen, {Pin Yu} and Bo Li and Dawn Song",
year = "2019",
month = "1",
day = "1",
language = "English (US)",
note = "7th International Conference on Learning Representations, ICLR 2019 ; Conference date: 06-05-2019 Through 09-05-2019",

}

TY - CONF

T1 - Characterizing audio adversarial examples using temporal dependency

AU - Yang, Zhuolin

AU - Chen, Pin Yu

AU - Li, Bo

AU - Song, Dawn

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Recent studies have highlighted adversarial examples as a ubiquitous threat to different neural network models and many downstream applications. Nonetheless, as unique data properties have inspired distinct and powerful learning principles, this paper aims to explore their potentials towards mitigating adversarial inputs. In particular, our results reveal the importance of using the temporal dependency in audio data to gain discriminate power against adversarial examples. Tested on the automatic speech recognition (ASR) tasks and three recent audio adversarial attacks, we find that (i) input transformation developed from image adversarial defense provides limited robustness improvement and is subtle to advanced attacks; (ii) temporal dependency can be exploited to gain discriminative power against audio adversarial examples and is resistant to adaptive attacks considered in our experiments. Our results not only show promising means of improving the robustness of ASR systems but also offer novel insights in exploiting domain-specific data properties to mitigate negative effects of adversarial examples.

AB - Recent studies have highlighted adversarial examples as a ubiquitous threat to different neural network models and many downstream applications. Nonetheless, as unique data properties have inspired distinct and powerful learning principles, this paper aims to explore their potentials towards mitigating adversarial inputs. In particular, our results reveal the importance of using the temporal dependency in audio data to gain discriminate power against adversarial examples. Tested on the automatic speech recognition (ASR) tasks and three recent audio adversarial attacks, we find that (i) input transformation developed from image adversarial defense provides limited robustness improvement and is subtle to advanced attacks; (ii) temporal dependency can be exploited to gain discriminative power against audio adversarial examples and is resistant to adaptive attacks considered in our experiments. Our results not only show promising means of improving the robustness of ASR systems but also offer novel insights in exploiting domain-specific data properties to mitigate negative effects of adversarial examples.

UR - http://www.scopus.com/inward/record.url?scp=85071178031&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071178031&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85071178031

ER -