Main content

Incorporating Anomaly Detection Techniques Within SPEA-2

Show full item record

Title: Incorporating Anomaly Detection Techniques Within SPEA-2
Author: Venkatesan, Tamizhselvan
Department: School of Computer Science
Program: Computer Science
Advisor: Wineberg, Mark
Abstract: Incorporating anomaly detection techniques within SPEA-2 Evolutionary Multi-Objective Optimization utilizes evolutionary algorithms to obtain optimal solutions involving multiple objectives, which often are conflicting in nature. This optimal solution set is known as a Pareto set and the corresponding fitness for the Pareto set is the Pareto front. Multi-Objective evolutionary algorithms can produce solution sets whose evaluations closely approach those of the Pareto front. This degree of closeness is called convergence. Multi-Objective Optimization algorithm often involves a trade-off between the convergence of the algorithm and the diversity of the solutions formed. Regular MOEA’s use a variety of diversity measures to try to control this trade-off. However, an alternative concept to diversity has been introduced in the field of data-mining, that of anomalous points. For our research we found preserving anomalous points, which have two important characteristics, namely sparsity and dissimilarity with respect to the other points in the dataset, improved the overall search direction and selection process of the evolutionary algorithm. In this thesis, we will investigate the importance of anomalous points within EMOA, which runs counter to how it's being used within the field of data-mining, through its incorporation within a Multi-Objective evolutionary algorithm framework known as SPEA-2.
URI: http://hdl.handle.net/10214/17498
Date: 2019-10
Rights: Attribution-NonCommercial-ShareAlike 4.0 International
Terms of Use: All items in the Atrium are protected by copyright with all rights reserved unless otherwise indicated.


Files in this item

Files Size Format View
Venkatesan_Tamizhselvan_201910_Msc.pdf 1.553Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record

Attribution-NonCommercial-ShareAlike 4.0 International Except where otherwise noted, this item's license is described as Attribution-NonCommercial-ShareAlike 4.0 International