This announcement is being cross-posted to several newsgroups.
Please, excuse the duplication.
Dear All,
The following paper is now available=20
as HTM Pages:
http://www.iephb.ru/~spirov/te-tech/te-tech.htm
and as ZIP-compressed
PostScript file via anonymous ftp from:
ftp://ftp.iephb.ru/pub/spirov/te_tech.zip (528 KB)
These materials are submitted to "Evol.Comp.Journal".=20
Title: "VARIABLE-LENGTH GENOMES, OPERATORS OF TRANSPOSITION AND =
PARASITIC MOBILE GENETIC ELEMENTS IN SIMULATIONS OF EVOLUTION"
Authors: Alexander V. Spirov and Alexander B. Kazansky
ABSTRACT
The progress in the field of evolutionary self-learning algorithms is =
primarily associated with the development of the set of operators, =
realising the process of evolutionary search and adaptation. It is a =
matter of fact, that the low "evolvability" of neo-darwinian population =
with the classical set of evolutionary operators put a fundamental =
limitation on the effectivity and universality of evolutionary =
computations, based on this type of evolving system. The realisation of =
this fact stimulates the constant interest of specialists in the field =
of evolutionary computations to the achievements in evolutionary and =
molecular biology. Godsends of nature can be adapted and applied to =
evolutionary algorithms for perfection of the instrument for solving of =
the hard optimisation problems. The creative role of viruses and =
virus-like elements in evolution attracts nowadays a great interest of =
evolutionary biologists as well as specialists in evolutionary =
computations.
In this work we put forward a new approach to evolutionary algorithms - =
Transposable Elements (TE) technique. This technique was evaluated and =
tested by the example of solving of biological, as well as classical =
test behavioural problems (the John Muir's trail in ant's navigation =
problem).
The approach is based on the application of new algorithms, modelling =
co-evolution of host population and a set of selfish/parasite mobile =
genetic elements (transposons). These elements have ability to insertion =
in the host's genome and to move from one genome site to another. =
Besides, transposons are heritable and can be transmitted to the next =
generations as well as to the other host during 'contacts'.
Our TE technique presupposes the application of special new TE operators =
of evolutionary computations along with classical ones. The TE operators =
act on genes-strings of variable length just as known variable-length =
operators of duplication, elimination and random addition.
The TE technique is effective in cases, when traditional neo-darwinian =
populations in framework of classical approaches loses the evolvability. =
This technique gives selection advantage to functionally redundant, more =
complex programs in the bottlenecks of evolutionary process.
The developed method was successfully applied to the simulation of =
biological evolution and to solving of ant's navigation test for =
non-standard variant of trail. The possibilities for application of =
proposed technique for solving GA-hard problems has being discussed as =
well.
KEYWORDS: self-organisation, evolution, transposons, genetic algorithms, =
genetic networks, artificial life, evolutionary biology.
(13 pages, 9 color illus.)
PS: The PS-file was generated by WinWord 7 and tested
by GhostScript viewer for Windows.=20
________________________
Alexander V.Spirov (PhD)
The Sechenov Institute of Evolutionary Physiology & Biochemistry, Thorez =
Pr. 44, S.-Petersburg, 194223, Russia
phone/fax +7 (812)552 3219; fax + 7 (812)552-3012 =
http://www.iephb.ru/~spirov/; Email spirov at iephb.ru
and
Institute for High-Performance Computing & Data Bases, P.O. Box 71, =
St.Petersburg 194291, RUSSIA
