Semantic Scholar Open Access 2021 10 sitasi

Investigating Feature Selection and Explainability for COVID-19 Diagnostics from Cough Sounds

F. Avila A. H. Poorjam Deepak Mittal Charles Dognin Ananya Muguli +4 lainnya

Abstrak

In this paper, we propose an approach to automatically classify COVID-19 and non-COVID-19 cough samples based on the combination of both feature engineering and deep learning models. In the feature engineering approach, we develop a support vector machine classifier over high dimensional (6373D) space of acoustic features. In the deep learning-based approach, on the other hand, we apply a convolutional neural network trained on the log-mel spectrograms. These two methodologically diverse models are then combined by fusing the probability scores of the models. The proposed system, which ranked 9th on the 2021 Diagnosing COVID-19 using Acoustics (Di- COVA) challenge leaderboard, obtained an area under the receiver operating characteristic curve (AUC) of 0:81 on the blind test data set, which is a 10:9% absolute improvement compared to the baseline. Moreover, we analyze the explainability of the deep learning-based model when detecting COVID-19 from cough signals. Copyright © 2021 ISCA.

Topik & Kata Kunci

Penulis (9)

F

F. Avila

A

A. H. Poorjam

D

Deepak Mittal

C

Charles Dognin

A

Ananya Muguli

R

Rohit Kumar

S

Srikanth Raj Chetupalli

S

Sriram Ganapathy

M

M. Singh

Format Sitasi

Avila, F., Poorjam, A.H., Mittal, D., Dognin, C., Muguli, A., Kumar, R. et al. (2021). Investigating Feature Selection and Explainability for COVID-19 Diagnostics from Cough Sounds. https://doi.org/10.21437/interspeech.2021-2197

Akses Cepat

Informasi Jurnal
Tahun Terbit
2021
Bahasa
en
Total Sitasi
10×
Sumber Database
Semantic Scholar
DOI
10.21437/interspeech.2021-2197
Akses
Open Access ✓