marvin-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE : [Marvin-devel] [LIBNN] reports


From: Matthieu Lagacherie
Subject: RE : [Marvin-devel] [LIBNN] reports
Date: Thu, 31 Jul 2003 15:11:20 +0200


)> -----Message d'origine-----
)> De : address@hidden [mailto:marvin-
)> address@hidden De la part de Olivier
)> Ricordeau
)> Envoyé : jeudi 31 juillet 2003 14:32
)> À : address@hidden
)> Objet : Re: [Marvin-devel] [LIBNN] reports
)> 
)> -----BEGIN PGP SIGNED MESSAGE-----
)> Hash: SHA1
)> 
)> Matthieu Lagacherie wrote:
)> > Hi!
)> > Olivier can you write a little report on the enhanced
backpropagation ?
)> > I'll write soon a report on the weight decay, you can use the
pattern
)> > used for the backpropagation report on your ftp.
)> > See you!
)> 
)> I've added this to my todo list. It won't be done soon, but it will
be
)> done. I still hope I'll have the time to code XML save/export before
I
)> go on vacation.
)> 
)> O.

A little question: the functions designed to deal with XML would be
coded 
with a visitor ?
I've tested the weight decay propagation, it effective now :).
This kind of learning is designed to improve generalization, I've do
somes 
little tests and... it's functioning well :)

)> 
)> - --
)> - -= *Olivier RICORDEAU* =-        http://freefolks.org
)>           < LibNN addict http://libnn.org >
)> lynx -source http://freefolks.org/key.asc | gpg --import
)> address@hidden _Might the source be with you_
)> -----BEGIN PGP SIGNATURE-----
)> Version: GnuPG v1.2.2 (GNU/Linux)
)> 
)> iD8DBQE/KQwjDFRYKP6DJEwRAmquAKCrzj6ItKMgBLG9qQrmD6DX+FHWBgCgs/tp
)> aqgo3LhP/lxhLDa8itYPMSQ=
)> =HpbQ
)> -----END PGP SIGNATURE-----
)> 
)> 
)> 
)> http://mail.nongnu.org/mailman/listinfo/marvin-devel
)> http://savannah.nongnu.org/projects/marvin/

* Matthieu LAGACHERIE - 0688197361 - EPITA Promo 2005 - 3IE *
* LibNN addict <www.libnn.org>, <www.marvinproject.org>     *
* GPG Key : www.epita.fr/~lagach_m/key.asc                  *
* fp : 6BC5 FC26 8E60 6D50 D253  9052 E835 378F 6FE5 E574   *





reply via email to

[Prev in Thread] Current Thread [Next in Thread]