Ecological theory provides insights about evolutionary computation
- Published
- Accepted
- Subject Areas
- Computational Biology, Adaptive and Self-Organizing Systems, Data Mining and Machine Learning, Theory and Formal Methods
- Keywords
- eco-evolutionary dynamics, ecological theory, evolutionary computation theory, multidisciplinary synthesis, diversity maintenance
- Copyright
- © 2018 Dolson et al.
- Licence
- This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ Preprints) and either DOI or URL of the article must be cited.
- Cite this article
- 2018. Ecological theory provides insights about evolutionary computation. PeerJ Preprints 6:e27315v1 https://doi.org/10.7287/peerj.preprints.27315v1
Abstract
Evolutionary algorithms often incorporate ecological concepts to help maintain diverse populations and drive continued innovation. However, while there is strong evidence for the value of ecological dynamics, a lack of overarching theoretical framework renders the precise mechanisms behind these results unclear. These gaps in our understanding make it challenging to predict which approaches will be most appropriate for a given problem. Biologists have been developing ecological theory for decades, but the resulting body of work has yet to be translated into an evolutionary computation context. This paper lays the groundwork for such a translation by applying ecological theory to three different selection mechanisms in evolutionary computation: fitness sharing, lexicase selection, and Eco-EA. First, we use ecological ideas to establish a framework that clarifies how these selection schemes are alike and how they differ. We then build upon this framework by using metrics from ecology to gather empirical data about the underlying differences in the population dynamics that these approaches produce. Specifically, we measure interaction networks and phylogenetic diversity within the population to explore long-term stable coexistence. Notably, we find that selection methods affect phylogenetic diversity differently than phenotypic diversity. These results can inform parameter selection, choice of selection scheme, and the development of new selection schemes.
Author Comment
An extended abstract based on this work was presented at the GECCO 2018 conference. The full version is being submitted to IEEE Transactions on Evolutionary Computation.