MSc Thesis Proposal Announcement by Kumaran Ragunathan: "Link Prediction based on Subgraphs"

Tuesday, September 17, 2019 - 09:30 to 11:30

SCHOOL OF COMPUTER SCIENCE

 

The School of Computer Science at the University of Windsor is pleased to present …

 

MSc Thesis Proposal by:   Kumaran Ragunathan
 
Date:  Tuesday 17th September 2019
Time:  9:30 am – 11:30am
Location: Lambton Tower Room 3105
 

Abstract: 

Link prediction is considered as a key research area in Social Network Analysis. There are various types of social networks existing such as friendship and air traffic network. Many methods have been proposed to predict links such as heuristic methods including common neighbors, Adamic Adar and Katz, as well as some machine learning methods including SVM, bagging, and deep learning. Even though heuristic methods perform well on certain types of a social network, they have lack accuracy for other types. Finding the best heuristic or combination of heuristics is a challenging task. Recent researches, WLNM and SEAL, show that the heuristics can be learned automatically by extracting subgraphs for each target link. However, those methods have several drawbacks. To overcome those shortcomings, we introduce a new framework that includes a new subgraph extracting and a node labeling algorithms. Then we use a Convolutional Neural Network to predict future links. We expected to outperform the heuristic methods as well as the state-of-the-art subgraphing methods.
 

Thesis Committee:

Internal Reader: Dr. Jianguo Lu
External Reader: Dr. Abdulkadir Hussein
Advisor: Dr.Ziad Kobti

 

Thesis Proposal Announcement

 

5113 Lambton Tower, 401 Sunset Ave., Windsor ON, N9B 3P4 519-253-3000 Ext. 3716 csgradinfo@uwindsor.ca

(519)253-3000