Properties of Genetic Representations of Neural Architectures

Balakrishnan, K. and Honavar, V. (1995) Properties of Genetic Representations of Neural Architectures. Technical Report TR95-13, Department of Computer Science, Iowa State University.

Abstract

Properties of Genetic Representations of
Neural Architectures
Karthik Balakrishnan and Vasant Honavar
Department of Computer Science
Iowa State University
Ames, IA 50011. U.S.A
Abstract
Genetic algorithms and related evolutionary techniques offer a
promising approach for automatically exploring
the design space of neural architectures for artificial intelligence
and cognitive modeling. Central to this process of evolutionary design
of neural architectures (EDNA) is the choice of the representation scheme
that is used to encode a neural architecture in the form of a gene string
(genotype) and to decode a genotype into the corresponding neural
architecture (phenotype). The representation scheme used
not only constrains the class of neural architectures that are
representable (evolvable) in the system, but also determines  the
efficiency and the time-space complexity of the evolutionary design
procedure as a whole. This paper identifies and discusses a set of
properties that can be used to characterize different representations
used in EDNA and to design or select representations with the necessary
properties for particular classes of applications.


Subjects:
All uncategorized technical reports


ID code:
00000103
Deposited by:
Staff Account on 07 July 1995





Contact site administrator at: ssg@cs.iastate.edu