TY - GEN
T1 - Adversarial Examples for Improving End-to-end Attention-based Small-footprint Keyword Spotting
AU - Wang, Xiong
AU - Sun, Sining
AU - Shan, Changhao
AU - Hou, Jingyong
AU - Xie, Lei
AU - Li, Shen
AU - Lei, Xin
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/5
Y1 - 2019/5
N2 - In this paper, we explore the use of adversarial examples for improving a neural network based keyword spotting (KWS) system. Specially, in our system, an effective and small-footprint attention-based neural network model is used. Adversarial example is defined as a misclassified example by a model, but it is only slightly skewed from the original correctly-classified one. In the KWS task, it is a natural idea to regard the false alarmed or false rejected queries as some kind of adversarial examples. In our work, given a well-trained attention-based KWS model, we first generate adversarial examples using the fast gradient sign method (FGSM) and find that these examples can dramatically degrade the KWS performance. Using these adversarial examples as augmented data to retrain the KWS model, we finally achieve 45.6% relative and false reject rate (FRR) reduction at 1.0 false alarm rate (FAR) per hour on a collected dataset from a smart speaker.
AB - In this paper, we explore the use of adversarial examples for improving a neural network based keyword spotting (KWS) system. Specially, in our system, an effective and small-footprint attention-based neural network model is used. Adversarial example is defined as a misclassified example by a model, but it is only slightly skewed from the original correctly-classified one. In the KWS task, it is a natural idea to regard the false alarmed or false rejected queries as some kind of adversarial examples. In our work, given a well-trained attention-based KWS model, we first generate adversarial examples using the fast gradient sign method (FGSM) and find that these examples can dramatically degrade the KWS performance. Using these adversarial examples as augmented data to retrain the KWS model, we finally achieve 45.6% relative and false reject rate (FRR) reduction at 1.0 false alarm rate (FAR) per hour on a collected dataset from a smart speaker.
KW - adversarial examples
KW - attention
KW - end-to-end
KW - KWS
UR - http://www.scopus.com/inward/record.url?scp=85068970254&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2019.8683479
DO - 10.1109/ICASSP.2019.8683479
M3 - 会议稿件
AN - SCOPUS:85068970254
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 6366
EP - 6370
BT - 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019
Y2 - 12 May 2019 through 17 May 2019
ER -