Adversarial Example Detection by Classification for Deep Speech Recognition

Saeid Samizade, Zheng-Hua Tan, Chao Shen, Guan Xiaohong

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

1 Citationer (Scopus)

Abstrakt

Machine Learning systems are vulnerable to adversarial attacks and will highly likely produce incorrect outputs under these attacks. There are white-box and black-box attacks regarding to adversary’s access level to the victim learning algorithm. To defend the learning systems from these attacks, existing methods in the speech domain focus on modifying input signals and testing the behaviours of speech recognizers. We, however, formulate the defense as a classification problem and present a strategy for systematically generating adversarial example datasets: one for white-box attacks and one for black-box attacks, containing both adversarial and normal examples. The white-box attack is a gradient-based method on Baidu DeepSpeech with the Mozilla Common Voice database while the black-box attack is a gradient-free method on a deep model-based keyword spotting system with the Google Speech Command dataset. The generated datasets are used to train a proposed Convolutional Neural Network (CNN), together with cepstral features, to detect adversarial examples. Experimental results show that, it is possible to accurately distinct between adversarial and normal examples for known attacks, in both single-condition and multi-condition training settings, while the performance degrades dramatically for unknown attacks. The adversarial datasets and the source code are made publicly available.
OriginalsprogEngelsk
TitelICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Antal sider5
ForlagIEEE
Publikationsdato9 apr. 2020
Sider3102-3106
Artikelnummer9054750
ISBN (Trykt)978-1-5090-6632-2
ISBN (Elektronisk)978-1-5090-6631-5
DOI
StatusUdgivet - 9 apr. 2020
BegivenhedICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) - Barcelona, Spanien
Varighed: 4 maj 20208 maj 2020

Konference

KonferenceICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
LandSpanien
ByBarcelona
Periode04/05/202008/05/2020
NavnICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN1520-6149

Fingeraftryk Dyk ned i forskningsemnerne om 'Adversarial Example Detection by Classification for Deep Speech Recognition'. Sammen danner de et unikt fingeraftryk.

Citationsformater