Please use this identifier to cite or link to this item:
https://hdl.handle.net/2440/120900
Type: | Thesis |
Title: | Algorithms and Machine Learning for Evolving Images |
Author: | Neumann, Aneta |
Issue Date: | 2019 |
School/Discipline: | School of Computer Science |
Abstract: | Evolutionary algorithms have been used in a wide range of areas to discover novel solutions. The primary aim in the research area of evolutionary algorithms and art is to evolve artistic and creative outputs through an evolutionary process. In this thesis, we investigate how algorithms and machine learning methods can assist in evolving images. We begin by designing different mutation operators based on uniform, biased random and quasi-random walks, and subsequently study how their combination with a baseline mutation operator can lead to interesting image transition processes in terms of visual effects and artistic features. Motivated by our investigations of evolutionary image transition, we present an approach for the composition of new images from existing ones, that retain some salient features of the original images. We use evolutionary algorithms to create new images based on a fitness function that incorporates feature covariance matrices associated with different parts of the images. Furthermore, evolutionary algorithms can be employed to obtain a diverse set of solutions and this field has gained increasing attention over recent years. Diversity optimization in terms of features on the underlying problem allows one to obtain a better understanding of possible solutions to the problem at hand and can be utilised for algorithm selection when concerned with combinatorial optimization problems. We introduce discrepancy-based diversity optimization approaches for evolving diverse sets of images. Furthermore, we propose a new approach for evolutionary diversity optimization based on well-established multi-objective performance indicators. It bridges the areas of evolutionary diversity optimization and evolutionary multi-objective optimization and shows how techniques developed in evolutionary multi-objective optimization can be used to produce diverse sets of solutions of high quality for a given single-objective problem. |
Advisor: | Alexander, Bradley Michalewicz, Zbigniew |
Dissertation Note: | Thesis (Ph.D.) -- University of Adelaide, School of Computer Science, 2019 |
Keywords: | Artificial Intelligence Machine Learning Evolutionary Algorithms |
Provenance: | This electronic version is made publicly available by the University of Adelaide in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material you wish to be removed from this electronic version, please complete the take down form located at: http://www.adelaide.edu.au/legals |
Appears in Collections: | Research Theses |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Neumann2019_PhD.pdf | 126.69 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.