Jellyfish???
Peter Riddle
Hi there,
People in this news group talk about Jelly fish, what is it and what
does this have to do with backgammon.
--
Peter Riddle
E Mail Pe...@rydell.demon.co.uk
*******************************
John S Mamoun
Peter Riddle (Pe...@rydell.demon.co.uk) wrote:
: Hi there,
Jellyfish: a thin, translucent Cnidarian of the Scyphozoa class,
conspicuous in the medusa stage with an ironically sessile and
miniscule polyp. For some odd reason, it is also the brand name of
one of the world's strongest neural net Backgammon
programs--which is probably why you couldn't guess what it was.
In recent years it has been found that computers are sometimes better
than humans at playing this game, and players often get excited when
a "bot" makes a play that they would never think of yet might be
correct--thus changing how we conceive of the game, which is exciting.
Philosophers are currently debating the symbolic significance of
"jellyfish" as a label for a neural net program. Perhaps the name is
meant to symbolize the beauty and flexibility of the program's brain,
just as a Cnidarian Scyphozoan is beautiful and flexible. But the
jellyfish stings! Does JF sting? Perhaps, when it demonstrates how
dumb we are at this game. But it too is often dumb! What does that
mean? And isn't a computer program run by a silicon brain, which is
stiff? But Scyphozoans are wavy and soft. What about tentacles?
How do they figure sybolically into our conception of a "neural net?"
The debate rages on.
David Hamilton
In article <595c30$p...@netnews.upenn.edu>,
John S Mamoun <js...@mail2.sas.upenn.edu> wrote:
>Peter Riddle (Pe...@rydell.demon.co.uk) wrote:
>
>: Hi there,
>
>: People in this news group talk about Jelly fish, what is it and what
>: does this have to do with backgammon.
>: --
>: Peter Riddle
>: E Mail Pe...@rydell.demon.co.uk
>: *******************************
>
> Philosophers are currently debating the symbolic significance of
>"jellyfish" as a label for a neural net program. Perhaps the name is
>meant to symbolize the beauty and flexibility of the program's brain,
>just as a Cnidarian Scyphozoan is beautiful and flexible. But the
>jellyfish stings! Does JF sting? Perhaps, when it demonstrates how
>dumb we are at this game. But it too is often dumb! What does that
>mean? And isn't a computer program run by a silicon brain, which is
>stiff? But Scyphozoans are wavy and soft. What about tentacles?
>How do they figure sybolically into our conception of a "neural net?"
>The debate rages on.
I always assumed that it was because the jellyfish's cephalization is
commonly called a neural net (or nerve net).
John S Mamoun
David Hamilton (dham...@wam.umd.edu) wrote:
: In article <595c30$p...@netnews.upenn.edu>,
Cephalization is not a trend evident in the jellyfish; cephalization
is mainly associated with higher forms of life, particularly mammals
and vertebrates in general. A nerve net exists in all Cnidarians (my
Biology text emphasizes the Hydra as the nerve net type organism).
But a nerve net means there is essentially no cephalization, since the
net is not segmented and differentiated enough to classify as such.
Cephalization is therefore associated with a type of neural organization,
just as a nerve net is a type of neural organization. Cephalization
tends towards a more specialized analytical system (in the brain), with
greater complexity in each specialized compartment, and the evolution of a
large ganglion cluster (which is also what the brain is).
But there is a problem here. The computer program Jellyfish is a highly
specialized silicon "ganglion," focused on understanding in its own way
only one way of thinking, namely the Backgammon logic. A jellyfish neural
net, however, integrates information in a very general way, with little
specialization towards any sort of logic. Thus, although both are
semantically known as "neural nets," they are functionally non-analogous.
In fact, considering that the Jellyfish program's analytical
engine is totally differentiated (focusing on one and only one logic)
and is extremely complex in understanding that logic, it is more analogous
to a squid ganglion than a jellyfish neural net. The squid ganglion, at
least, does have components that are both specialized and relatively complex.
Therefore, a more suitable name for the computer program is not "Jellyfish,"
but rather, "Squid."
Peter Riddle
In article <596jfa$i...@rac10.wam.umd.edu>, David Hamilton
<dham...@wam.umd.edu> writes
>I always assumed that it was because the jellyfish's cephalization is
>commonly called a neural net (or nerve net).
Fredrik Dahl
The name of JellyFish:
When I gave it this name I considered the following:
- compared to a biological neural net it has a very low number
of 'brain cells', and not being a biologist, I figured it might
be comparable to the jellyfish.
- Though primitive, the jellyfish is a very successful creature,
or so I believe. It's a survivor.
- I thought there were all to many not-so-strong programs with
names that promised too much, like 'super-champion-pro-backgammon',
and thought it would be fun to see them all eaten by a lower life form.
- I liked the sound of the word.
The JellyFish Homepage, btw, is:
http://www.effect.no/jelly.htm
Fredrik Dahl.