The Robotics Institute

Carnegie Mellon Robotics Institute

Copyright notice: This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Abstract

In evolutionary search, the term evolvability as defined in [Altenberg 94] refers to ``the ability of a population to produce variants fitter than any yet existing''. In this paper, we examine a few existing mechanisms which provide the potential for the evolvability of a population to itself evolve. One key property that we identify among such mechanisms is a many-to-one genotype-to-phenotype mapping, which permits variations in evolvability to occur independent of fitness. Another is the propensity for individuals to become increasingly conservative in parent-offspring transmission as they become more fit, a phenomenon which becomes stronger as selection pressure becomes weaker.

Keywords

genetic algorithms, neural networks

Notes

Associated Center(s) / Consortia: Center for Integrated Manfacturing Decision Systems

Associated Lab(s) / Group(s): Evolutionary Computation

Text Reference

Matthew Glickman and Katia Sycara, "Comparing Mechanisms for Evolving Evolvability," Proceedings of 1999 Genetic and Evolutionary Computation Conference Workshop Program, July, 1999.

BibTeX Reference

@inproceedings{Glickman_1999_3235,
   author = "Matthew Glickman and Katia Sycara",
   editor = "A. Wu",
   title = "Comparing Mechanisms for Evolving Evolvability",
   booktitle = "Proceedings of 1999 Genetic and Evolutionary Computation Conference Workshop Program",
   month = "July",
   year = "1999",
}

The Robotics Institute

Menu