报告摘要:Molecular identification and discovery play a pivotal role in biochemical analysis, environmental governance, customs inspection, and other critical fields. Spectroscopic instruments, including mass spectrometry, infrared spectroscopy, nuclear magnetic resonance (NMR) spectroscopy, and X-ray spectroscopy, are fundamental tools for exploring the microscopic molecular world. However, current spectroscopic data analysis methods for molecular identification and discovery primarily rely on database searching and human professional expertise. These approaches face inherent limitations: they struggle to identify and discover novel molecules beyond existing databases, while suffering from low analysis efficiency, poor accuracy, and high costs. To address these challenges, we have conducted systematic research, exploration, and system development across multiple dimensions: data collection and cleaning, the development of a family of spectroscopic large foundation models, the creation of AI Agents for spectroscopic analysis, algorithm design grounded in biochemical principles, and practical application deployment in biomedicine and environmental protection. In this talk, I will elaborate on the research framework of SpectraAI and its latest progress, while highlighting promising research directions for future.
讲者简介:夏俊现为香港科技大学(广州)、香港科技大学联署助理教授和博士生导师。博士毕业于浙江大学-西湖大学联合培养项目计算机科学与技术专业,导师为讲席教授李子青(Stan Z. Li, IEEE Fellow); 博士期间在中国电信人工智能研究院 TeleAI担任见习研究员,导师为李学龙教授。申请人在与本项目相关的研究工作中有着丰富的积累,取得了国际领先的科研成果, 在Nature Methods、ICML、NeurIPS、ICLR等顶级期刊会议发表50余篇论文,其中以一作和通讯身份发表31篇,一篇一作论文入选WWW 2022 Most Influential Papers。主持首批国家自然科学基金青年学生基础研究项目(博士生)、首届中国电子学会-腾讯博士生科研激励项目(全国17项)等,参与首批科技部新一代人工智能国家重大科技专项“首席科学家负责制”试点项目(全国 3 项)等国家重大战略科技项目,担任NeurIPS、ICLR、KDD、IJCAI等AI顶级会议的领域主席或高级程序委员会委员,担任Nature Communications、TPAMI等期刊审稿人。曾获Rising Star in AI by KAUST、DAAD AINet Fellowship、Apple AI/ML Scholar Finalist、西湖大学校长奖章、浙江省优秀毕业生、国家奖学金等。