Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Ge, Shijia, Zhang, Weixiang, Xie, Shuzhao, Yan, Baixu, Wang, Zhi
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2501.00064
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915086701953024
author Ge, Shijia
Zhang, Weixiang
Xie, Shuzhao
Yan, Baixu
Wang, Zhi
author_facet Ge, Shijia
Zhang, Weixiang
Xie, Shuzhao
Yan, Baixu
Wang, Zhi
contents Respiratory sound classification plays a pivotal role in diagnosing respiratory diseases. While deep learning models have shown success with various respiratory sound datasets, our experiments indicate that models trained on one dataset often fail to generalize effectively to others, mainly due to data collection and annotation \emph{inconsistencies}. To address this limitation, we introduce \emph{Lungmix}, a novel data augmentation technique inspired by Mixup. Lungmix generates augmented data by blending waveforms using loudness and random masks while interpolating labels based on their semantic meaning, helping the model learn more generalized representations. Comprehensive evaluations across three datasets, namely ICBHI, SPR, and HF, demonstrate that Lungmix significantly enhances model generalization to unseen data. In particular, Lungmix boosts the 4-class classification score by up to 3.55\%, achieving performance comparable to models trained directly on the target dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2501_00064
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Lungmix: A Mixup-Based Strategy for Generalization in Respiratory Sound Classification
Ge, Shijia
Zhang, Weixiang
Xie, Shuzhao
Yan, Baixu
Wang, Zhi
Sound
Machine Learning
Audio and Speech Processing
Respiratory sound classification plays a pivotal role in diagnosing respiratory diseases. While deep learning models have shown success with various respiratory sound datasets, our experiments indicate that models trained on one dataset often fail to generalize effectively to others, mainly due to data collection and annotation \emph{inconsistencies}. To address this limitation, we introduce \emph{Lungmix}, a novel data augmentation technique inspired by Mixup. Lungmix generates augmented data by blending waveforms using loudness and random masks while interpolating labels based on their semantic meaning, helping the model learn more generalized representations. Comprehensive evaluations across three datasets, namely ICBHI, SPR, and HF, demonstrate that Lungmix significantly enhances model generalization to unseen data. In particular, Lungmix boosts the 4-class classification score by up to 3.55\%, achieving performance comparable to models trained directly on the target dataset.
title Lungmix: A Mixup-Based Strategy for Generalization in Respiratory Sound Classification
topic Sound
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2501.00064