Year: 2021
Type: dissertation
Abstract: Optimization is a process to find an optimized value of a mathematical function and it can be obtained by swarm intelligence algorithms.Clustering is grouping data with similar characteristics to different groups so that one group can differ from another based-on characteristic.This thesis includes the study of traditional clustering algorithm such as K-means (KM).Also, it includes the study of swarm intelligence algorithms such as Firefly Algorithm (FA) and Cuckoo Search Algorithm (CSA).BioColour data is used as input of these algorithms for comparison.The Euclidean distance (ED) cost function is used for all algorithms to minimize the error rate and find the best results.First, clustering results are obtained from the KM algorithm including graphical representation and cluster centroids.Secondly, the original data sample is used as input of FA and CSA to obtain the results.Then cluster centroids are used as input of FA and CSA.The comparison results show that if we use data directly without clustering centroids as input of FA and CSA then CSA shows better results.If we use clustering centroids as input of FA and CSA, then FA shows better results.Comparison of KM, CSA, and FA shows that KM is not reducing the distance as CSA and FA.It means cluster centroids can improve the efficiency of the FA if we compare it with the CSA.Lastly, the execution time complexity comparison is done, and results show that CSA completes the processing earlier than FA.
Author: Kamran Jabbar
Institution: University of Eastern Finland
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: Helsingin Yliopisto
OpenAlex ID: W3215733761
