Thursday, December 17, 2020 - 15:00 to 17:00
SCHOOL OF COMPUTER SCIENCE
The School of Computer Science is pleased to present…
MSc Thesis Defense by: Nimish Verma
Date: Thursday December 17th 2020
Time: 3:00-5:00 pm
Zoom URL: https://zoom.us/j/99188628287?
Reminder: If interested in attending this thesis event, contact the Graduate Secretary at csgradinfo@uwindsor.ca for the passcode.
Abstract:
This thesis proposes new meta-heuristic optimization frameworks for single and bi-objective problems. These frameworks have a hierarchical multi-population structure that is inspired by the pack structure observed in grey wolves. The populations are called Alpha, Beta, Delta, and Omega; and are divided in such a way that the individuals in Alpha are better than Beta, which are better than Delta, the remaining individuals are in Omega Population. We also propose a local search algorithm which is executed in each population and evolves that local population independently. This algorithm is capable of inputting external individuals called as guide leaders. The evolution process runs the local search in Alpha, Beta, and Delta population, and then one guide leader from each population is sent to guide the local search in Omega population. Additionally, a mutation operator and elitism is incorporated to improve convergence rates and exploration ability.
The proposed frameworks are tested on IEEE’s Congress of Evolutionary Computation (CEC) benchmarks for unconstrained real-parameter optimization in single and bi-objective problems. This thesis also addresses the defects that are studied in the traditional grey wolf algorithms and performs a comparison of proposed framework with the state of the art methods.
Keywords: meta-heuristic, optimization, grey wolves, multi-population
Thesis Committee:
Internal Reader: Dr. Saeed Samet
External Reader: Dr. Roozbeh Razavi-Far
Co-Advisors: Dr Pooya Moradian Zadeh, Dr. Ziad Kobti
Chair: Dr. Kalyani Selvarajah
MSc Thesis Defense Announcement
5113 Lambton Tower 401 Sunset Ave. Windsor ON, N9B 3P4 (519) 253-3000 Ext. 3716 csgradinfo@uwindsor.ca