[{"data":1,"prerenderedAt":377},["ShallowReactive",2],{"publication-2024\u002Fmulti-view-spectrogram-transformer-for-respiratory-sound-classification-en":3,"publication-members":65},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"_hidden":6,"authors":10,"authors_orcid":16,"year":22,"doi":23,"openalex_id":24,"venue":19,"abstract_screenshot":19,"keywords":25,"body":34,"_type":58,"_id":59,"_source":60,"_file":61,"_stem":62,"_extension":63,"locale":64},"\u002Fpublications\u002F2024\u002Fmulti-view-spectrogram-transformer-for-respiratory-sound-classification","2024",false,"","Multi-View Spectrogram Transformer for Respiratory Sound Classification","Deep neural networks have been applied to audio spectrograms for respiratory sound classification. Existing models often treat the spectrogram as a synthetic image while overlooking its physical characteristics. In this paper, a Multi-View Spectrogram Transformer (MVST) is proposed to embed different views of time-frequency characteristics into the vision transformer. Specifically, the proposed MVST splits the mel-spectrogram into different-sized patches, representing the multi-view acoustic elements of a respiratory sound. The patches and positional embeddings are fed into transformer encoders to extract the attentional information among patches through a self-attention mechanism. Finally, a gated fusion scheme is designed to automatically weigh the multi-view features to highlight the best one in a specific scenario. Experimental results on the ICBHI dataset demonstrate that the MVST significantly outperforms state-of-the-art methods for classifying respiratory sounds. The code is available at: https:\u002F\u002Fgithub.com\u002Fwentaoheunnc\u002FMVST.",[11,12,13,14,15],"He, Wentao","Yan, Yuchen","Ren, Jianfeng","Bai, Ruibin","Jiang, Xudong",[17,18,19,20,21],"0000-0002-6319-1639","0000-0003-3359-4012",null,"0000-0003-1722-568X","0000-0002-9104-2315",2024,"https:\u002F\u002Fdoi.org\u002F10.1109\u002Ficassp48485.2024.10445825","W4392903171",[26,27,28,29,30,31,32,33],"Spectrogram","Computer science","Transformer","Encoder","Speech recognition","Artificial intelligence","Pattern recognition (psychology)","Engineering",{"type":35,"children":36,"toc":55},"root",[37],{"type":38,"tag":39,"props":40,"children":41},"element","p",{},[42,45,53],{"type":43,"value":44},"text","Deep neural networks have been applied to audio spectrograms for respiratory sound classification. Existing models often treat the spectrogram as a synthetic image while overlooking its physical characteristics. In this paper, a Multi-View Spectrogram Transformer (MVST) is proposed to embed different views of time-frequency characteristics into the vision transformer. Specifically, the proposed MVST splits the mel-spectrogram into different-sized patches, representing the multi-view acoustic elements of a respiratory sound. The patches and positional embeddings are fed into transformer encoders to extract the attentional information among patches through a self-attention mechanism. Finally, a gated fusion scheme is designed to automatically weigh the multi-view features to highlight the best one in a specific scenario. Experimental results on the ICBHI dataset demonstrate that the MVST significantly outperforms state-of-the-art methods for classifying respiratory sounds. The code is available at: ",{"type":38,"tag":46,"props":47,"children":51},"a",{"href":48,"rel":49},"https:\u002F\u002Fgithub.com\u002Fwentaoheunnc\u002FMVST",[50],"nofollow",[52],{"type":43,"value":48},{"type":43,"value":54},".",{"title":7,"searchDepth":56,"depth":56,"links":57},2,[],"markdown","content:publications:2024:multi-view-spectrogram-transformer-for-respiratory-sound-classification.md","content","publications\u002F2024\u002Fmulti-view-spectrogram-transformer-for-respiratory-sound-classification.md","publications\u002F2024\u002Fmulti-view-spectrogram-transformer-for-respiratory-sound-classification","md","en",[66,79,84,95,102,110,116,125,132,138,143,153,160,169,175,187,196,205,211,219,224,232,238,246,250,260,267,275,280,288,294,302,308,316,321,327,337,345,351,359,364,372],{"_path":67,"title":68,"name":69,"role":70,"email":19,"image":71,"category":72,"interests":73,"order":56,"_id":78},"\u002Fmembers\u002Fstaff\u002Falain-chong","Vice President for Global Affairs and Partnerships · Professor of Information Systems and Digital Innovation","Alain Chong","Deputy Director of Lab","assets\u002F8.png","staff",[74,75,76,77],"信息系统与运作管理","计算机科学与运筹学","Information Systems and Operations Management","Computer Science and Operations Research","content:members:staff:alain-chong.md",{"_path":67,"title":80,"role":81,"interests":82,"_id":83},"全球事务与合作副校长 · 信息系统与数字创新教授","实验室副主任",[74,75],"content:members:staff:alain-chong.zh-CN.md",{"_path":85,"title":86,"name":87,"role":88,"email":19,"image":89,"category":72,"interests":90,"order":93,"_id":94},"\u002Fmembers\u002Fstaff\u002Fanthony-belloti","Professor","Anthony Belloti","Core Member","assets\u002F41.png",[91,92],"Machine Learning and Credit Risk Model","Model Risks",9,"content:members:staff:anthony-belloti.md",{"_path":85,"title":96,"role":97,"interests":98,"_id":101},"计算机科学系教授","核心成员",[99,100],"机器学习与信用风险模型","模型风险","content:members:staff:anthony-belloti.zh-CN.md",{"_path":103,"title":104,"name":104,"role":88,"email":19,"image":105,"category":72,"interests":106,"order":108,"_id":109},"\u002Fmembers\u002Fstaff\u002Fboon-giin-lee","Boon Giin Lee","assets\u002F31.jpg",[107],"Intelligent Sensor and Extended Reality",11,"content:members:staff:boon-giin-lee.md",{"_path":103,"title":111,"role":97,"interests":112,"_id":115},"人机交互实验室负责人 · 计算机科学系副教授",[113,114],"人机交互 HCI","智能传感与扩展现实技术","content:members:staff:boon-giin-lee.zh-CN.md",{"_path":117,"title":118,"name":118,"role":119,"email":19,"image":120,"category":72,"interests":121,"order":123,"_id":124},"\u002Fmembers\u002Fstaff\u002Fcong-cao","Cong Cao","Direction Leader","assets\u002FCC.png",[122],"Science and technology policy and institutional reform",7,"content:members:staff:cong-cao.md",{"_path":117,"title":126,"name":127,"role":128,"interests":129,"_id":131},"宁波诺丁汉大学商学院创新学教授","曹聪","方向带头人",[130],"科技政策与体制改革","content:members:staff:cong-cao.zh-CN.md",{"_path":133,"title":134,"name":134,"role":88,"email":19,"image":135,"category":72,"order":136,"_id":137},"\u002Fmembers\u002Fstaff\u002Fdave-towey","Dave Towey","assets\u002F32.jpg",8,"content:members:staff:dave-towey.md",{"_path":133,"title":139,"role":97,"interests":140,"_id":142},"计算机科学系教授 · 计算机科学系主任",[141],"计算机科学与语言学","content:members:staff:dave-towey.zh-CN.md",{"_path":144,"title":145,"name":145,"role":88,"email":19,"image":146,"category":72,"interests":147,"order":151,"_id":152},"\u002Fmembers\u002Fstaff\u002Ffazl-ullah-khan","Fazl Ullah Khan","assets\u002F44.png",[148,149,150],"Computer Network","Computer Architecture and Network Security","Software Engineering",12,"content:members:staff:fazl-ullah-khan.md",{"_path":144,"title":154,"role":97,"interests":155,"_id":159},"计算机科学系助理教授 · IEEE 高级会员",[156,157,158],"计算机网络","计算机和网络安全","软件工程","content:members:staff:fazl-ullah-khan.zh-CN.md",{"_path":161,"title":162,"name":163,"role":88,"email":19,"image":164,"category":72,"interests":165,"order":167,"_id":168},"\u002Fmembers\u002Fstaff\u002Fheng-yu","Associate Professor","Heng Yu","assets\u002FHENGYU.png",[166],"Embedded Systems Design",17,"content:members:staff:heng-yu.md",{"_path":161,"title":170,"name":171,"role":97,"interests":172,"_id":174},"计算机科学系副教授","于恒",[173],"嵌入式系统设计","content:members:staff:heng-yu.zh-CN.md",{"_path":176,"title":162,"name":177,"role":88,"email":19,"image":178,"category":72,"interests":179,"order":185,"_id":186},"\u002Fmembers\u002Fstaff\u002Fheshan-du","Heshan Du","assets\u002Fhesahndu.png",[180,181,182,183,184],"Logic, Knowledge Representation and Reasoning","Geographic Information Systems","Operations Research","Machine Learning","Reinforcement Learning",20,"content:members:staff:heshan-du.md",{"_path":176,"title":170,"name":188,"role":97,"interests":189,"_id":195},"杜何珊",[190,191,192,193,194],"逻辑与知识表示","地理信息系统","运筹学","机器学习","强化学习","content:members:staff:heshan-du.zh-CN.md",{"_path":197,"title":198,"name":199,"role":88,"email":19,"image":200,"category":72,"interests":201,"order":203,"_id":204},"\u002Fmembers\u002Fstaff\u002Fhuan-jin","Assistant Professor","Huan Jin","assets\u002Fhuanjin.png",[202,183],"Optimisation",21,"content:members:staff:huan-jin.md",{"_path":197,"title":206,"name":207,"role":97,"interests":208,"_id":210},"计算机科学系助理教授","靳欢",[209,193],"优化","content:members:staff:huan-jin.zh-CN.md",{"_path":212,"title":162,"name":213,"role":119,"email":19,"image":214,"category":72,"interests":215,"order":217,"_id":218},"\u002Fmembers\u002Fstaff\u002Fjianfeng-ren","Jianfeng Ren","assets\u002F42.jpg",[183,216],"Computer Vision",3,"content:members:staff:jianfeng-ren.md",{"_path":212,"title":170,"name":220,"role":128,"interests":221,"_id":223},"任剑锋",[193,222],"计算机视觉","content:members:staff:jianfeng-ren.zh-CN.md",{"_path":225,"title":226,"name":226,"role":119,"email":19,"image":227,"category":72,"interests":228,"order":230,"_id":231},"\u002Fmembers\u002Fstaff\u002Fjiawei-li","Jiawei Li","assets\u002F11.png",[229],"Computer Science and Artificial Intelligence",15,"content:members:staff:jiawei-li.md",{"_path":225,"title":233,"name":234,"role":128,"interests":235,"_id":237},"计算机科学系助理教授 · 英国诺丁汉大学博士后","李家炜",[236],"计算机与人工智能","content:members:staff:jiawei-li.zh-CN.md",{"_path":239,"title":162,"name":240,"role":88,"email":19,"image":241,"category":72,"interests":242,"order":244,"_id":245},"\u002Fmembers\u002Fstaff\u002Fmatthew-pike","Matthew Pike","assets\u002F43.jpg",[243],"Digitalised Learning",16,"content:members:staff:matthew-pike.md",{"_path":239,"title":170,"role":97,"interests":247,"_id":249},[248],"数字化学习","content:members:staff:matthew-pike.zh-CN.md",{"_path":251,"title":198,"name":252,"role":88,"email":19,"image":253,"category":72,"interests":254,"order":258,"_id":259},"\u002Fmembers\u002Fstaff\u002Fning-xue","Ning Xue","\u002Fimages\u002Fuon-logo.png",[255,256,257],"Artificial Intelligence","Computational Intelligence","Combinatorial Optimization",13,"content:members:staff:ning-xue.md",{"_path":251,"title":206,"name":261,"role":97,"interests":262,"_id":266},"薛宁",[263,264,265],"人工智能","计算智能","组合优化","content:members:staff:ning-xue.zh-CN.md",{"_path":268,"title":198,"name":269,"role":88,"email":19,"image":270,"category":72,"interests":271,"order":273,"_id":274},"\u002Fmembers\u002Fstaff\u002Fqian-zhang","Qian Zhang","assets\u002Fqz.png",[272,216,183],"Image Processing",14,"content:members:staff:qian-zhang.md",{"_path":268,"title":206,"name":276,"role":97,"interests":277,"_id":279},"张茜",[278,222,193],"图像处理","content:members:staff:qian-zhang.zh-CN.md",{"_path":281,"title":86,"name":282,"role":283,"email":19,"image":284,"category":72,"interests":285,"orcid":20,"order":286,"_id":287},"\u002Fmembers\u002Fstaff\u002Fruibin-bai","Ruibin Bai","Director of Lab","assets\u002F38.png",[77],1,"content:members:staff:ruibin-bai.md",{"_path":281,"title":289,"name":290,"role":291,"interests":292,"_id":293},"教授","白瑞斌","实验室主任",[75],"content:members:staff:ruibin-bai.zh-CN.md",{"_path":295,"title":296,"name":296,"role":119,"email":19,"image":297,"category":72,"interests":298,"order":300,"_id":301},"\u002Fmembers\u002Fstaff\u002Fsean-he","Sean He","assets\u002F39.png",[216,299,183],"Data Analytics",5,"content:members:staff:sean-he.md",{"_path":295,"title":303,"name":304,"role":128,"interests":305,"_id":307},"计算机科学系教授 · 国家级讲席学者","何祥健",[222,306,193],"数据分析","content:members:staff:sean-he.zh-CN.md",{"_path":309,"title":310,"name":310,"role":88,"email":19,"image":311,"category":72,"interests":312,"order":314,"_id":315},"\u002Fmembers\u002Fstaff\u002Ftianxiang-cui","Tianxiang Cui","assets\u002Ftianxiangcui.png",[256,313,183,184],"Operation Research",19,"content:members:staff:tianxiang-cui.md",{"_path":309,"title":154,"name":317,"role":97,"interests":318,"_id":320},"崔天翔",[264,319,193,194],"运筹研究","content:members:staff:tianxiang-cui.zh-CN.md",{"_path":322,"title":86,"name":323,"role":88,"email":19,"image":324,"category":72,"order":325,"_id":326},"\u002Fmembers\u002Fstaff\u002Fxiuping-hua","Xiuping Hua","assets\u002FxiupignHua.png",10,"content:members:staff:xiuping-hua.md",{"_path":322,"title":328,"name":329,"role":97,"interests":330,"_id":336},"金融、会计与经济系教授","华秀萍",[331,332,333,334,335,99],"资产定价","公司金融","衍生品","金融科技","创新金融和普惠金融","content:members:staff:xiuping-hua.zh-CN.md",{"_path":338,"title":86,"name":339,"role":88,"email":19,"image":340,"category":72,"interests":341,"order":343,"_id":344},"\u002Fmembers\u002Fstaff\u002Fying-weng","Ying Weng","assets\u002Fyingweng.png",[216,272,342],"IoT",4,"content:members:staff:ying-weng.md",{"_path":338,"title":96,"name":346,"role":97,"interests":347,"_id":350},"翁莹",[222,278,348,349],"物联网 IoT","无线网络安全与服务质量","content:members:staff:ying-weng.zh-CN.md",{"_path":352,"title":198,"name":353,"role":88,"email":19,"image":354,"category":72,"interests":355,"order":357,"_id":358},"\u002Fmembers\u002Fstaff\u002Fyuan-yao","Yuan Yao","assets\u002Fyuanyao.png",[356],"Autonomous Agents and Multi-Agent Systems",18,"content:members:staff:yuan-yao.md",{"_path":352,"title":206,"name":360,"role":97,"interests":361,"_id":363},"姚远",[362],"自主智能体与多智能体系统","content:members:staff:yuan-yao.zh-CN.md",{"_path":365,"title":198,"name":366,"role":119,"email":19,"image":367,"category":72,"interests":368,"order":370,"_id":371},"\u002Fmembers\u002Fstaff\u002Fzheng-lu","Zheng Lu","assets\u002F13.png",[369],"Computer Science",6,"content:members:staff:zheng-lu.md",{"_path":365,"title":206,"name":373,"role":128,"interests":374,"_id":376},"卢正",[375],"计算机科学","content:members:staff:zheng-lu.zh-CN.md",1782639662220]