Abstract
We present an alternative to the cellular encoding
technique Gruau 1992 for evolving graph and network
structures via genetic programming. The new technique,
called edge encoding, uses edge operators rather than
the node operators of cellular encoding. While both
cellular encoding and edge encoding can produce all
possible graphs, the two encodings bias the genetic
search process in different ways; each may therefore be
most useful for a different set of problems. The
problems for which these techniques may be used, and
for which we think edge encoding may be particularly
useful, include the evolution of recurrent neural
networks, finite automata, and graph-based queries to
symbolic knowledge bases. In this preliminary report we
present a technical description of edge encoding and an
initial comparison to cellular encoding. Experimental
investigation of the relative merits of these encoding
schemes is currently in progress.
Users
Please
log in to take part in the discussion (add own reviews or comments).