Evolution of Analog Networks using Local String Alignment on Highly Reorganizable Genomes

• Presented at: NASA/DoD Conference on Evolvable Hardware (EH'2004), Seattle, 24-26 June
• Published in: Proceedings of the NASA/DoD Conference on Evolvable Hardware (EH'2004), p. 30-37
• : , 2004

We introduce and apply a genetic representation for analog electronic circuits based on the association of character strings extracted from the genome with the ter-minals and parameters of components, and the use of local string alignment to generate the connection between components. The representation produces a variable ge-nome length structure that tolerates the execution of ma-jor genome reorganization operators such as duplication and transposition, along with less disruptive ones such as character insertion, deletion and substitution. The repre-sentation can be applied also to other analog networks such as artificial neural networks, control systems, and biological genetic regulatory networks.

Note: Zebulum, R.S. et al. (eds)

Keywords: AGE ; analog genetic encoding ; neuroevolution ; neuromodulation ; evolution of analog electronic circuits ; reinforcement learning ; implicit encoding ; implicit genetic encoding

Reference

• LIS-CONF-2004-002
• View record in Web of Science

• URL: http://ehw.jpl.nasa.gov/events/nasaeh04/

Record created on 2006-01-12, modified on 2012-03-20