讲座题目:Finding Outstanding Aspects and Contrast Subspaces
主讲人: 裴 健 教授
主持人: 林学民 教授
开始时间:2014-11-17(周一上午)10:00-11:30
讲座地址:中北校区数学馆201室
主办单位:软件学院、校科技处
报告人简介:
Jian Pei is Canada Research Chair (Tier 1) in Big Data Science and a Professor of Computing
Science at Simon Fraser University. He is veteran of data mining research and his work
has been embraced by industry and government. Since 2000, his research has focused on
developing effective and efficient ways to analyze and capitalize on the vast stores of
data housed in applications such as social networks, network security informatics, healthcare
informatics, business intelligence, and web searches. A prolific and widelycited author,
Professor Pei has received several prestigious awards including induction as a Fellow of IEEE.
报告内容摘要:
In our recent endeavor of computational health informatics/intelligence, we face a series of
interesting problems of finding outstanding aspects that distinguish a specific object from its peers.
In this talk, I will introduce the unsupervised and supervised versions. Specifically, given a set of objects and a
query object, all in a multidimensional space, the unsupervised version finds the minimal subspaces
where the object is most outlying against the other objects in the set. The supervised version assumes
class labels (either positive or negative) for objects in the set and the query object, and finds the
minimal subspaces where the query object is most dissimilar to the other objects in the same class and
similar to those in the other class. I will discuss the subtle differences betwen this group of problems
and the traditional outlier detection problem.Furthermore, I will review our preliminary progress and demonstrate
the challenges remained open.