Proposing effective coordinate search methods for solving large-scale expensive black-box optimization problems

In engineering and science, optimization plays a vital role in many real-world applications. In this work, several novel optimization algorithms based on Coordinate Search (CS) algorithm are proposed. CS is a gradient-free technique and we have enhanced them for solving Black-box, non-convex, and expensive large-scale problems. These CS-based algorithms can handle mixed-type variables. When an optimization problem is large-scale and expensive, it is a very challenging problem to solve because it is intersecting two conflicting properties. Large-scale problems require extensive fitness evaluations, but each evaluation is time consuming. It gets more challenging when the budget is limited, which is the case in most real-word applications. The proposed CS-based algorithms reduce the search space exponentially; this makes it a powerful method for optimizing high-dimensional problems with limited budget. The proposed algorithms show a very promising performance on optimizing high-dimensional problems; tested on the CEC-2013 benchmarks problems and neural network training.

Name thesis:degree_name

Master of Applied Science (MASc)

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2020

Author dc:creator

• Rokhsatyazdi, Ehsan

Advisor dc:contributor.advisor

• Rahnamayan, Shahriyar

• Coordinate-search
• Gradient-free
• Non-convex
• Neural-network
• Large-scale optimization

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/1234

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/1234