Neural Network based Chessprogram
Michael Niemeck
Hi there.
I am currently implementing a chess programm based on neural networks.
Has anyone ever done this. If so, what were your experiences? I hope
to have some results in summer. Anyone interested in this?
If any more detail is wanted in this newsgroup, please say so.
I really don't know is this will ever work, I am doing this mainly to
find out. It is just my perosnal feeling that a neural network should
be good in handling chess positions...well, let's see.
I have also spent a lot of time at the underlying library of the chess
rules. How vain was my work? Is there such a library already
available, if so, where.
I have also done an interface that I can have my program play against
gnuchess (And thus train it).
Without starting a flame war, how 'good' is gnuchess? I just want to
know if there is any sense in using gnuchess, or if there is another
program in the UNIX world mor suitable for the beginning.
If you want any more information, please sayso, I will post it then
here on this newsgroup.
Michael.
--
______ __
__(___ / __ /
( / /\/ / MICHAEL NIEMECK
/|/ /| / Technical University of Vienna
/ | / |\/
(______ /_/|/ |/_
(________/ / _|_ e922...@stud1.tuwien.ac.at
Kevin Barnes
>Hi there.
>I am currently implementing a chess programm based on neural networks.
>Has anyone ever done this. If so, what were your experiences? I hope
>to have some results in summer. Anyone interested in this?
>If any more detail is wanted in this newsgroup, please say so.
>I really don't know is this will ever work, I am doing this mainly to
>find out. It is just my perosnal feeling that a neural network should
>be good in handling chess positions...well, let's see.
>I have also spent a lot of time at the underlying library of the chess
>rules. How vain was my work? Is there such a library already
>available, if so, where.
>I have also done an interface that I can have my program play against
>gnuchess (And thus train it).
>Without starting a flame war, how 'good' is gnuchess? I just want to
>know if there is any sense in using gnuchess, or if there is another
>program in the UNIX world mor suitable for the beginning.
Try Crafty (ftp willis.cis.uab.edu directory pub/hyatt), but be
warned, if you train against a computer expect to play like one. It's
true for this carbon based life form and it'll be true for your neural
net too.
kba...@t1sys.cfer.com
(303) 633-3689
Jay Scott
>I am currently implementing a chess programm based on neural networks.
>Has anyone ever done this.
I have info about a couple of neural net chess programs at
my Machine Learning in Games web pages:
http://forum.swarthmore.edu/~jay/learn-game/index.html
See SAL and NeuroChess.
Jay Scott
j...@forum.swarthmore.edu
Martin Borriss
You might want to contact the author of the program below (Neuron played
on GICS). Although it looks like it wasn't a big success ;)
Martin
Statistics for Neuron (Last disconnected Fri Aug 04, 20:33 MET DST):
rating RD win loss draw total best
Blitz 1631 102.2 4 10 0 14
Standard 1475 212.6 0 2 0 2
Lightning ---- 350.0 0 0 0 0
Wild ---- 350.0 0 0 0 0
Bughouse ---- 350.0 0 0 0 0
Atomic ---- 350.0 0 0 0 0
1: I am "nessi version 0.3", code written by Udo Sprute.
2:
3: I have been compiled with GNU C++ 2.6.3 with full optimization.
4: I am working with an ICS interface front-end written in perl.
5: You can send comments or questions to jma...@cck.uni-kl.de.
6: Aborts will always be accepted.
7:
8: WARNING: THIS IS STILL EXPERIMENTAL. En-passant does not work yet.
9: USE abort IN CASE OF PROBLEMS.
In article <4kvto9$6...@news.tuwien.ac.at>, e922...@stud1.tuwien.ac.at (Michael Niemeck) writes:
>
> Hi there.
> I am currently implementing a chess programm based on neural networks.
> Has anyone ever done this. If so, what were your experiences? I hope
> to have some results in summer. Anyone interested in this?
> If any more detail is wanted in this newsgroup, please say so.
> I really don't know is this will ever work, I am doing this mainly to
> find out. It is just my perosnal feeling that a neural network should
> be good in handling chess positions...well, let's see.