drogoul@lpia4.Berkeley.EDU (Drogoul Alexis 42.28.96.38) (12/13/90)
Hello! I am going to work on modelling some activities of an anthill using reactive agents (under a system called Eco-Problem-Solving). I would like to have all the references on this subject (Did somebody use Distributed Artificial Intelligence to model eco-systems like anthills ? What are the other works on "artificial ethology" ?). Thanks a lot. Alexis DROGOUL LAFORIA - Universite PARIS VI 75232 PARIS CEDEX 05 FRANCE
fostel@eos.ncsu.edu (Gary Fostel) (12/21/90)
Ants fall into a catagory that is probably heavily populated in the
world of small scale biology. A great deal of their behavior is
controlled by what seem to be simple environmental cues (e.g. the
elevation of the ground to avoid wasting energy going uphill) and by
a set of temporary cues depositied by other ants as pheromes. The
ants follow gradients of certain pheromes and the collective action
of the ants following the steady accumulation of the pheromes in the
environment together with the facts of the environments layout lead
astonishingly enough to copnvergence on a small set of similar observable
behaviors. There is a rough anaolgy to neural systems here that is
probably closest in very young brains that are still growing connections
based on some simple envrionmental cues.
The model is, as above, more pervasis then simpy ants and neurons since
something like it is at work in specialization during development in
general as some cells decide to produce the proteins needed to build a
liver and some for a lung based -- at least in part -- on the relative
concentration of some biochemical marking agents. This has been shown
pretty conclusively in the development of fruit fly larvea anyway. The
conditions needed for this rather loose collection of constraints to
converge towards well defined structures from such an ergodic soup is
quite amazing to me ... but mother nature seems to have the technique
well in hand.
But on the subject of ants and learning and AI : there's gold in them
that hills. Wish you luck digging it out.
----GaryFostel---- Department of Computer Science
North Carolina State Universitynewsgst@sequoia.execu.com (Net News) (12/23/90)
There was a guy at the MIT media lab (I thought was
working under Minsky) that was using ants to demonstate
some of the principles (there were principles? don't
hold me as affirming that) in his society of mind stuff.
All I remember is seeing a demo sometime when I was haunting
the halls over there (fun!). The little ants (it was graphic)
were modelled as far as food finding and trail making behaviour
was concerned. Someone might have a better idea on what was
happening and I have no idea where the ant guy's work ended.
E.O.Wilson's new book on ants is supposed to be really neat.
That about exhausting my knowledge on that topic.
jj
newsgst@execu.com
jjoy@lucy.wellesley.edu