Abstract
Recently, the question of whether artifacts obtained from a generative art system can be judged as creative based on the characteristics of their offspring has received considerable attention. Here, we focus on the question of whether aesthetic images can be evolved by considering characteristics of their offspring. We introduce a formal model for designing fitness functions for use in automated evolution of aesthetic images whereby genotypes are evaluated relative to certain characteristics of their offspring. We describe the results of an experiment using such an indirect fitness scheme that promotes offspring diversity in order to help select for parent phenotypes with desired symmetry and complexity properties. We use as our image generation platform a variant of the Sims' classical Evolving Expressions generative art system. © 2014 Springer-Verlag.
Cite
CITATION STYLE
Greenfield, G. (2014). An indirect fitness scheme for automated evolution of aesthetic images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8601 LNCS, pp. 85–94). Springer Verlag. https://doi.org/10.1007/978-3-662-44335-4_8
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
