- author = "Clinton Jon Woodward",
- title = "Ecosystems, Complexity, Topology and Evolutionary Computation",
- school = "Swinburne University of Technology",
- year = "2010",
- address = "Australia",
- keywords = "genetic algorithms, genetic programming, ESEC, python",
-
URL = "
https://researchbank.swinburne.edu.au/file/9f3f293b-d5a8-4c95-88b5-c73eee5bcf9b/1/Clinton%20Jon%20Woodward%20Thesis.pdf",
- size = "497 pages",
-
abstract = "Evolutionary algorithms have been applied to an
increasing range of complex problem domains. A
challenge for many applications is the discovery of
appropriate structures and processes that allow
solutions, and solution components, to emerge
efficiently.
The motivation of this thesis was to create a new ecosystem model of evolutionary computation (ESEC) and to investigate the influence that topology and interaction can have on the outcome of evolutionary search. The thesis begins by considering the field of ecology and models of ecosystems, with a particular emphasis on evolutionary models,structures and processes. Next, existing models of evolutionary computation are considered with a strong emphasis on aspects of topology. Modern developments in the field of graph theory provide new insight into complex systems and the properties of efficient structures.
A range of investigation themes have been developed for the ESEC model, and a detailed survey of topology models and properties was undertaken to guide the selection of suitable structures. An empirical study considers in detail the specific influence of various population structures on evolutionary search outcomes, and shows that the specification of population topology can influence both the efficacy and efficiency of evolutionary search.The results are a motivation for future investigations to consider in more detail how and why such influence can be used to an advantage as a way of optimising evolutionary search applications.",
- notes = "Supervisor: Tim Hendtlass",
