人机交互与虚拟现实中心学科前沿讨论班 通知(2016上半年NO.30) 打印 E-mail
作者:mudcup   
2016-04-12 14:10

山东大学计算机学院&软件学院

数媒中心&人机交互&虚拟现实研究中心

2015-2016学年 第二学期博、硕士研究生学科前沿讲座报告会第 3 次报告会 预告

一、报告时间: 2016年4月14日(周四)下午13:30分-15:30分

二、报告地点: 数字媒体教育部工程技术中心大楼111-113教室

三、出席人员:数媒中心&人机交互&虚拟现实中心 各年级相关方向的研究生和相关研究方向的教师

四、报告内容

报告1

Link Prediction by Exploiting Network Formation Games in Exchangeable Graph

报告人:王立强(博士生)

内容简介

Link prediction is an important research problem in social networks analysis aiming to study user behaviors and infer new interactions in the near future. Among the link prediction algorithms, the family of algorithms based on exchangeable graphs has been shown to be effective approaches recently. The network is firstly modeled as an exchangeable array, whose entries can be flexibly ’traced back to’ random function priors (e.g., block models, Gaussian Processes). However, the huge computational complexity inhibits its application to even moderate scale networks. In order to solve this problem, we present an online training algorithm (called ORFP) based on local Gaussian processes on sub-graphs. To further explore the sparsity of links in social networks, we also presents an improved algorithm based on network formation games. Moreover, the designed network formation games can also shed lights on the modeling of the ambiguity of the missing links (not observed or non-existing), which remains to be an open problem. Finally, We evaluate our method against state-of-the-art link prediction algorithms on real world datasets. The experimental results has demonstrated the effectiveness and efficiency of the proposed method for link prediction.

 

欢迎对报告感兴趣的其他老师和学生参加

 

 

联系我们: vr@sdu.edu.cn tel:(0531)88395759 山东大学计算机科学与技术学院人机交互与虚拟现实实验室 版权所有