行健讲坛学术讲座

第379期

时间:   2019年4月10日（周三）上午9：30

地点:   校本部东区翔英大楼 T516室

讲座:   基于人工智能的健康监护：从音频到自发身体活动领域的应用

          A.I.4Healthcare:  Applications from Audio to Spontaneous Physical  Activity

演讲者: 钱昆 特任研究员 日本东京大学

演讲者简介：钱昆博士，日本学术振兴会特别研究员（全球每年录用率10%），日本东京大学特任研究员，德国慕尼黑工业大学工学博士，主要研究人工智能与信号处理方向。钱博士与美国卡内基梅隆大学、英国帝国理工大学、新加坡南洋理工大学、中国科学院等国内外顶尖高校和科研机构保持合作关系，致力于深度学习在医疗健康、音频智能感知和数据挖掘方面的研究。目前以第一作者身份发表SCI收录期刊论文8篇，其中包括IEEE  Transactions on Biomedical Engineering、Annals of Biomedical Engineering、JASA等国际知名期刊，并在国际知名学术会议如ICASSP、EMBC、GlobalSIP上发表相关学术论文。钱博士长期担任国际知名期刊IEEE  Transactions on Cybernetics、IEEE  Transactions on Biomedical Engineering、IEEE  Transactions on Affective Computing、IEEE  Transactions on Neural Networks and Learning Systems、IEEE  Signal Processing Letters以及ICASSP、INTERSPEECH、EMBC、AVEC等国际知名学术会议审稿人。已授权中国发明类专利3项，德国专利1项。

讲座摘要：In  a traditional or classical A.I. paradigm, the human hand-crafted features are  extracted from the data by several signal processing methods, e.g., Fourier  transformation, wavelet transformation, empirical mode decomposition, etc.  Subsequently, a machine learning model can be trained when fed with those  features. Even though the performance and the robustness of the model could be  feasible for further implementations in real practice, the feature engineering  process, which needs specific domain knowledge, is still time-consuming, and  expensive. As an emerging technique, deep learning, can make it possible to make  models learn higher representations from the data itself. In  this presentation, Dr. Qian will present his main work in Technical University  of Munich, Germany, and his most recent work in The University of Tokyo. For his  work in Germany, the audio data can be used for diagnosing some diseases related  to the knowledge of body acoustics. For his work in Japan, the spontaneous  physical activity data can be good representations for screening the patients  suffering from the major depressive disorder.

邀请人：上海大学通信与信息工程学院 朱梦尧副教授

欢迎广大教师和学生参加！