There are various representations for encoding graph
structures, such as artificial neural networks (ANNs)
and circuits, each with its own strengths and
weaknesses. Here we analyse edge encodings and show
that they produce graphs with a node creation order
connectivity bias (NCOCB). Additionally, depending on
how input/ output (I/O) nodes are handled, it can be
difficult to generate ANNs with the correct number of
I/O nodes. We compare two edge encoding languages, one
which explicitly creates I/O nodes and one which
connects to pre-existing I/O nodes with parameterised
connection operators. Results from experiments show
that these parameterized operators greatly improve the
probability of creating and maintaining networks with
the correct number of I/O nodes, remove the
connectivity bias with I/O nodes and produce better
ANNs. These results suggest that evolution with a
representation which does not have the NCOCB will
produce better performing ANNs. Finally we close with a
discussion on which directions hold the most promise
for future work in developing better representations
for graph structures.
%0 Journal Article
%1 Hornby:2006:GPEM
%A Hornby, Gregory S.
%D 2006
%J Genetic Programming and Evolvable Machines
%K ANN CEEL, Circuits, Graphs, Neural PEEL, Representations, algorithms, genetic networks, programming,
%N 3
%P 231--252
%R doi:10.1007/s10710-006-9007-5
%T Shortcomings with using edge encodings to represent
graph structures
%U http://ic.arc.nasa.gov/publications/pdf/1212.pdf
%V 7
%X There are various representations for encoding graph
structures, such as artificial neural networks (ANNs)
and circuits, each with its own strengths and
weaknesses. Here we analyse edge encodings and show
that they produce graphs with a node creation order
connectivity bias (NCOCB). Additionally, depending on
how input/ output (I/O) nodes are handled, it can be
difficult to generate ANNs with the correct number of
I/O nodes. We compare two edge encoding languages, one
which explicitly creates I/O nodes and one which
connects to pre-existing I/O nodes with parameterised
connection operators. Results from experiments show
that these parameterized operators greatly improve the
probability of creating and maintaining networks with
the correct number of I/O nodes, remove the
connectivity bias with I/O nodes and produce better
ANNs. These results suggest that evolution with a
representation which does not have the NCOCB will
produce better performing ANNs. Finally we close with a
discussion on which directions hold the most promise
for future work in developing better representations
for graph structures.
@article{Hornby:2006:GPEM,
abstract = {There are various representations for encoding graph
structures, such as artificial neural networks (ANNs)
and circuits, each with its own strengths and
weaknesses. Here we analyse edge encodings and show
that they produce graphs with a node creation order
connectivity bias (NCOCB). Additionally, depending on
how input/ output (I/O) nodes are handled, it can be
difficult to generate ANNs with the correct number of
I/O nodes. We compare two edge encoding languages, one
which explicitly creates I/O nodes and one which
connects to pre-existing I/O nodes with parameterised
connection operators. Results from experiments show
that these parameterized operators greatly improve the
probability of creating and maintaining networks with
the correct number of I/O nodes, remove the
connectivity bias with I/O nodes and produce better
ANNs. These results suggest that evolution with a
representation which does not have the NCOCB will
produce better performing ANNs. Finally we close with a
discussion on which directions hold the most promise
for future work in developing better representations
for graph structures.},
added-at = {2008-06-19T17:35:00.000+0200},
author = {Hornby, Gregory S.},
biburl = {https://www.bibsonomy.org/bibtex/27766410391fab3aea5a2bef75a30c975/brazovayeye},
doi = {doi:10.1007/s10710-006-9007-5},
interhash = {1fac29b476cf11f9cffbacb34f94f1e0},
intrahash = {7766410391fab3aea5a2bef75a30c975},
issn = {1389-2576},
journal = {Genetic Programming and Evolvable Machines},
keywords = {ANN CEEL, Circuits, Graphs, Neural PEEL, Representations, algorithms, genetic networks, programming,},
month = {October},
notes = {3-parity. goal scoring robot},
number = 3,
pages = {231--252},
timestamp = {2008-06-19T17:41:46.000+0200},
title = {Shortcomings with using edge encodings to represent
graph structures},
url = {http://ic.arc.nasa.gov/publications/pdf/1212.pdf},
volume = 7,
year = 2006
}