Weighted Graph Compression using Genetic Algorithms

Networks are a great way to present information. It is easy to see how different objects interact with one another, and the nature of their interaction. However, living in the technological era has led to a massive surge in data. Consequently, it is very common for networks/graphs to be large. When graphs get too large, the computational power and time to process these networks gets expensive and inefficient. This is common in areas such as bioinformatics, epidemic contact tracing, social networks, and many others. Graph compression is the process of merging nodes that are highly connected into one super-node, thus shrinking the graph. The goal of graph compression is to merge nodes while mitigating the amount of information lost during the compression process. Unweighted graphs are largely studied in this area. However, in this thesis, we extend the approaches to compress weighted graphs via genetic algorithms and analyse the compression from an epidemic point of view. It is seen that edge weights provide vital information for graph compression. Not only this, but having meaningful edge weights is important as different weights can lead to different results. Moreover, both the original edge weights and adjusted edge weights produce different results when compared to a widely used community detection algorithm, the Louvain Algorithm. However, the different results may be helpful to public health officials. Lastly, the NSGA-II algorithm was implemented. It was found that NSGA-II is more suitable as a pre-processing tool, in order to find a target compression that introduces a comfortable level of distortion, and then using the single-objective genetic algorithm to achieve an improved solution for the target.

Name thesis:degree_name

M.Sc. Computer Science

Level thesis:degree_level

Masters

Discipline thesis:degree_discipline

Faculty of Mathematics and Science

Department dc:contributor.department

Department of Computer Science

Grantor

Brock University

Year dc:date.issued

2022

Author dc:creator

• Rutkowski, Emilia

• Genetic Algorithm
• Graph Compression
• Weighted Networks
• Contact Networks
• NSGA-II

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/10464/15768

OAI identifier oai:identifier

oai:brocku.scholaris.ca:10464/15768