学术讲座【Toward Machine Understanding】

07.07.2014  17:40

时间:2014年7月16日(周三)上午9:30

地点:仓山校区 成功楼603报告厅

主讲:英国Ulster大学 王晖教授

主办:数学与计算机科学学院、福建省网络安全与密码技术重点实验室

专家简介:王晖,现为英国Ulster大学教授。1985年、1988年在吉林大学计算机科学与技术学院获学士、硕士学位,1996年于英国Ulster大学获博士学位。1988年至1993年在吉林大学任教,1996年在英国Ulster大学获博士学位后留校任教,1996年至2002年在Ulster大学任讲师(lecturer),2002年至2006年任高级讲师(senior lecturer),2006年至2011年任副教授(reader),2011年至今任教授;2009年9月起任Ulster大学知识和数据工程研究室主任。长期从事机器学习,人工智能和知识工程,数据挖掘等领域的研究工作,他的很多原创性的工作,包括格子机学习方法、Neighbourhood Counting Kernel的相似度测量方法,在国际上得到认可并获得广泛关注。近5年独立承担欧盟基金项目(EU FP7) 3项(约80万英镑),英国投资北爱尔兰研发基金 Invest Northern Ireland R&D Programme(约35万英镑),英国皇家学会与中国自然科学基金等科研项目多项。近5年发表SCI 1区文章1篇,SCI 2区文章6篇,SCI 3区文章3篇。兼任国际期刊 《IEEE Transactions on SMC-B》、《International Journal of Machine Learning and Cybernetics》副主编,多个国际会议的大会共同主席。

报告摘要:Big Data is a hot term used to characterise today’s digital world. Data is being generated in large quantity and velocity, and data is complex due to variety – from structured (e.g. measurements), semi-structured (e.g. web pages), to unstructured (e.g. text, video). One of the challenges Big Data presents is how to get machines to understand data. If we could understand text, we could summarise it or answer questions. If we could understand video, we could detect events for surveillance purpose. Machine Understanding is an emerging field of study that is aimed at tackling the challenging problem of getting machines to understand data in a modern way, one that utilises recent advances in areas such as computational semantics and ontology, focuses on common `understanding’ operations that can be specialised for different types of data.
              In this talk, Professor Hui Wang will present his work on knowledge-supported learning as an approach to machine understanding. He will demonstrate a journey from Lattice Machine to Contextual Probability to Neighbourhood Counting Kernel, leading to a new approach to machine learning where learning from data is supported by existing knowledge. He will then present a case study on machine understanding in the context of a project.