Advancing Metadata-Convolutional Neural Networks with Multi-supervised Contrastive Learning and Metadata Insights for Respiratory Sound Analysis

  • Miao Liu
  • , Haojie Zhang
  • , Kun Qian*
  • , Bin Hu
  • , Toru Nakamura
  • , Taishin Nomura
  • , Jian Zhang
  • , Zhangguo Tang
  • , Björn W. Schuller
  • , Yoshiharu Yamamoto
  • , Huanzhou Li
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Metadata, as information describing data, encompasses an exhaustive description of various facets of the data. Despite the enhanced accuracy of existing respiratory sound detection methods through multifaceted research, these methods often under-utilise metadata. To explore the potential of metadata, we study its impact on detection performance. We adopt a multi-supervised contrastive learning approach and propose an improved Metadata-Convolutional Neural Network model for more effective extraction of metadata features. We use the International Conference in Biomedical Health Informatics (ICBHI) 2017 database for evaluation and achieve an average score of 59.48% on the official (6:4) split, surpassing current state-of-the-art methods. Moreover, utilising metadata increased the detection rate of respiratory sounds, with gender, a key predictive factor, outperforming other combinations when used with other metadata. Specifically, when combined with age, the average score reached 59.64%.

Original languageEnglish
Title of host publicationProceedings of the 11th Conference on Sound and Music Technology - Revised Selected Papers from CSMT 2024
EditorsKun Qian, Li Zhou, Qinglin Meng, Yongwei Gao
PublisherSpringer Science and Business Media Deutschland GmbH
Pages25-36
Number of pages12
ISBN (Print)9789819647828
DOIs
Publication statusPublished - 2025
Event11th National Conference on Sound and Music Technology, CSMT 2024 - Wuhan, China
Duration: 11 Oct 202413 Oct 2024

Publication series

NameLecture Notes in Electrical Engineering
Volume1404 LNEE
ISSN (Print)1876-1100
ISSN (Electronic)1876-1119

Conference

Conference11th National Conference on Sound and Music Technology, CSMT 2024
Country/TerritoryChina
CityWuhan
Period11/10/2413/10/24

Keywords

  • Metadata
  • Multi-supervised contrastive learning
  • Respiratory sound analysis

Fingerprint

Dive into the research topics of 'Advancing Metadata-Convolutional Neural Networks with Multi-supervised Contrastive Learning and Metadata Insights for Respiratory Sound Analysis'. Together they form a unique fingerprint.

Cite this