报告时间：12月30日上午10:30

报告地点：科技馆二楼扇形报告厅

报告题目:  Collaborative Filtering in Recommendation Systems with Medical  Applications

报告人：王子栋教授，英国布鲁奈尔大学

   王子栋博士现任英国布鲁奈尔大学工程设计与物理科学学院终身教授。王子栋教授于1996年获得德国洪堡基金，1998年获得日本科学促进会基金，2002年获得香港大学威廉蒙基金。多年来从事智能数据分析、生物信息学、控制理论、信号处理等方面的研究，在SCI刊物上发表国际论文五百余篇。现任或曾任十二种国际刊物的主编、副编辑或编委，包括Neurocomputing主编；国际系统科学杂志执行主编；IEEE自动控制汇刊副编缉；IEEE控制系统技术汇刊副编缉；IEEE神经网络汇刊副编缉； IEEE系统、人、与控制汇刊副编缉；IEEE信号处理汇刊副编缉等。曾任东华大学CJ讲座教授（2008-2010）,  清华大学国家特聘教授（2011-2014）。 现为IEEE Fellow，IEEE出版社编委，英国皇家统计学会理事，曾任旅英华人自动化及计算机协会主席。

Abstract:  In this talk, we discuss a novel user-based collaborative filtering (CF)  algorithm with improved performance for recommendation systems. The statistical  information set (SIS) of individual rating data is, for the first time, employed  to analyze the user rating habit, thereby facilitating the performance  improvement of the CF algorithm. On the basis of the SIS, a new yet  comprehensive similarity measure (SM) is proposed to quantify the distance  between two users with focus on both the users' preferences on items and their  rating habits. Compared with the traditional SM, our proposed SM is more general  with clearer application insights in complicated situations. The developed CF  algorithm makes full use of the known information of a recommendation system,  which merits high prediction accuracy and wide application potential. The  developed CF algorithm is applied to a real-world disease (Friedreich's ataxia)  assessment system, where both the effectiveness and the superiority of our  proposed algorithm are demonstrated.