A Hybrid Between TOA and Lévy Flight Trajectory for Solving Different Cluster Problems

Nagaraju Devarakonda, Ravi Kumar Saidala, Raviteja Kamarajugadda

Source Title: International Journal of Cognitive Informatics and Natural Intelligence (IJCINI)15(4)

ISSN: 1557-3958|EISSN: 1557-3966|EISBN13: 9781799859857|DOI: 10.4018/IJCINI.20211001.oa39

MLA

Devarakonda, Nagaraju, et al. "A Hybrid Between TOA and Lévy Flight Trajectory for Solving Different Cluster Problems." IJCINI vol.15, no.4 2021: pp.1-25. http://doi.org/10.4018/IJCINI.20211001.oa39

APA

Devarakonda, N., Saidala, R. K., & Kamarajugadda, R. (2021). A Hybrid Between TOA and Lévy Flight Trajectory for Solving Different Cluster Problems. International Journal of Cognitive Informatics and Natural Intelligence (IJCINI), 15(4), 1-25. http://doi.org/10.4018/IJCINI.20211001.oa39

Chicago

Devarakonda, Nagaraju, Ravi Kumar Saidala, and Raviteja Kamarajugadda. "A Hybrid Between TOA and Lévy Flight Trajectory for Solving Different Cluster Problems," International Journal of Cognitive Informatics and Natural Intelligence (IJCINI) 15, no.4: 1-25. http://doi.org/10.4018/IJCINI.20211001.oa39

Export Reference

Favorite Full-Issue Download

View Full Text HTML

View Full Text PDF

Abstract

In data analysis applications for extraction of useful knowledge, clustering plays an important role. The major shortcoming of traditional clustering algorithms is exhibiting poor performance in solving complex data cluster problems. This research paper introduces a novel hybrid optimization technique based clustering approach. This paper is designed with two main objectives: designing efficient function optimization algorithm and developing advanced data clustering approach. In achieving the first objective, the standard TOA is first enhanced by hybridizing with Lévy flight trajectory and benchmarked on 23 functions. A new clustering approach is developed by conjoining k-means algorithm and Lévy flight TOA. Tested the numerical complexity of the proposed novel clustering approach on 10 UCI clustering datasets and 4 web document cluster problems. Conducted several simulation experiments and done an analysis of the results. The obtained graphical and statistical analysis reveals that the proposed novel clustering approach yields better quality clusters.