Scholarlink

<p>This paper presents a novel cross-dataset transfer learning approach for cough-based COVID-19 detection, enhancing model performance through data augmentation. Our methodology significantly improves results compared to baseline methods. An ablation study highlights the importance of alpha mixup among various hyperparameters in optimizing performance. The final model achieves an unweighted accuracy of 88.19%. Additionally, we provide a comparative summary with previous studies on the same evaluation set to offer insights into cough-based detection methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Cross-dataset COVID-19 transfer learning with data augmentation

  • 期刊论文

  • 发表日期：

    2025年2月13日

查看全文
  • Bagus Tris Atmaja,
  • Zanjabila,
  • Suyanto,
  • Wiratno Argo Asmoro,
  • Akira Sasou

摘要

This paper presents a novel cross-dataset transfer learning approach for cough-based COVID-19 detection, enhancing model performance through data augmentation. Our methodology significantly improves results compared to baseline methods. An ablation study highlights the importance of alpha mixup among various hyperparameters in optimizing performance. The final model achieves an unweighted accuracy of 88.19%. Additionally, we provide a comparative summary with previous studies on the same evaluation set to offer insights into cough-based detection methods.