Unsupervised Adversarial Domain Adaptation Based On The Wasserstein Distance For Acoustic Scene Classification
Tutkimustuotos › › vertaisarvioitu
Standard
Unsupervised Adversarial Domain Adaptation Based On The Wasserstein Distance For Acoustic Scene Classification. / Drossos, Konstantinos; Magron, Paul; Virtanen, Tuomas.
2019 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA). IEEE, 2019. (IEEE Workshop on Applications of Signal Processing to Audio and Acoustics).Tutkimustuotos › › vertaisarvioitu
Harvard
APA
Vancouver
Author
Bibtex - Lataa
}
RIS (suitable for import to EndNote) - Lataa
TY - GEN
T1 - Unsupervised Adversarial Domain Adaptation Based On The Wasserstein Distance For Acoustic Scene Classification
AU - Drossos, Konstantinos
AU - Magron, Paul
AU - Virtanen, Tuomas
PY - 2019/10/22
Y1 - 2019/10/22
N2 - A challenging problem in deep learning-based machine listening field is the degradation of the performance when using data from unseen conditions. In this paper we focus on the acoustic scene classification (ASC) task and propose an adversarial deep learning method to allow adapting an acoustic scene classification system to deal with a new acoustic channel resulting from data captured with a different recording device. We build upon the theoretical model of HΔH-distance and previous adversarial discriminative deep learning method for ASC unsupervised domain adaptation, and we present an adversarial training based method using the Wasserstein distance. We improve the state-of-the-art mean accuracy on the data from the unseen conditions from 32% to 45%, using the TUT Acoustic Scenes dataset.
AB - A challenging problem in deep learning-based machine listening field is the degradation of the performance when using data from unseen conditions. In this paper we focus on the acoustic scene classification (ASC) task and propose an adversarial deep learning method to allow adapting an acoustic scene classification system to deal with a new acoustic channel resulting from data captured with a different recording device. We build upon the theoretical model of HΔH-distance and previous adversarial discriminative deep learning method for ASC unsupervised domain adaptation, and we present an adversarial training based method using the Wasserstein distance. We improve the state-of-the-art mean accuracy on the data from the unseen conditions from 32% to 45%, using the TUT Acoustic Scenes dataset.
KW - Acoustic scene classification
KW - unsupervised domain adaptation
KW - Wasserstein distance
KW - adversarial training
U2 - 10.1109/WASPAA.2019.8937231
DO - 10.1109/WASPAA.2019.8937231
M3 - Conference contribution
SN - 978-1-7281-1124-7
T3 - IEEE Workshop on Applications of Signal Processing to Audio and Acoustics
BT - 2019 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA)
PB - IEEE
ER -