[MuS-L] Vortrag von M.Boden: Donnerstag, 18.4.2002 (Artificial Life)

Andreas Schamanek schamanek at gmx.net
Thu Apr 18 00:24:47 CEST 2002


Werte MuS-Ln!

Heute, Donnerstag, 18. April 2002, findet im Rahmen des ALTENBERG
SEMINARS IN THEORETICAL BIOLOGY ('Model Building in Biology', Hoersaal
2, Biozentrum, Althanstrasse 14, 18 Uhr c.t.) der Vortrag

   Margaret BODEN: Artificial Life as Theoretical Biology

statt, was recht spannend werden koennte (siehe auch Abstract unten).

Artificial Life (AL) zaehlt zu den wesentlichsten modernen Ansaetzen
der Modellbildung und Simulation, hat Beruehrungspunkte und
ueberschneidet sich zum Teil mit Neuronalen Netzen, Kuenstlicher
Intelligenz und Multi-Agenten-Systemen. Dabei soll AL die biologischen
Vorlagen getreuer abbilden (als klassische Methoden), und ueberdies
verstaendlicher und einsichtiger. Der Erfolg der modernen Ansaetze
spricht -- sagen wir mal -- fuer sie. Wie auch immer, Margaret Boden
scheint auch einen guten Ueberblick ueber AL zu geben, worueber ich
mich sehr freue. Hier noch der Abstract zum Vortrag:


Abstract

Artificial life (A-Life) is a new form of mathematical biology. It
uses computer models and computational concepts to develop theories of
specific aspects of living organisms (e.g. flocking, cockroach
locomotion, and cricket phonotaxis) and to explore general aspects of
life such as replication, evolution, and self-organization. It also
aims to arrive at a definition of life in general: "Life as it could
be," not just "Life as we know it." A few A-life workers also accept a
fourth aim, although most do not: synthesizing (real) life in
computers, either as physical robots or as virtual reality,
informational "creatures" inside computer memory. The fourth aim is
highly controversial, and involves several tricky philosophical
questions. 

In this talk, I'll ignore it and focus instead on methodological
issues. A-Life methods fall into seven classes, all focussed on
bottom-up emergence, or self-organization. They are: 

(1) "Once and for all" programs (i.e., not evolutionary); 
(2) "Once and for all" robots; 
(3) Evolutionary programs; 
(4) Evolutionary robotics; 
(5) Studies of the computational properties of cellular automata;
(6) Evolutionary hardware; and 
(7) Computational neuroethology. 

The last method is especially close to the biological (neurological
and behavioral) data. But all are inspired by biological data and/or
theory.


--------------------------------------------------------------------

So sei's,

-- Andreas 






More information about the MuS-L mailing list