N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. Homepage Information Progress Downloads Mailing lists Contact
N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks. N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks.N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks.N.A.R.I.A. is an open project about creating artificial intelligence with the help of recurrent asymmetric neural networks.

N.A.R.I.A.

Artificial Intelligence: Recurrent Asymmetric Networks

Progress


This page is intended to keep people updated with the current status of the project. It will also serve as a sort of history scrollback for the important steps of the N.A.R.I.A. project.

01 June 2002: Posted by Olivier Reisch (doctomoe@naria.karasuma.net)

After a very long time of inactivity, an updated version of the N.A.R.I.A. Library source code has been released. Changes are relatively minor, though they allow the network to be a lot more scalable now. The test app has been modified to make use of the ability and accuracy meters now allow to see how effective the network stays in it's task given the number of neurons you attribute it. There have also been some minor bug fixes.
As an interesting sidenote, i have started coding a new test app for the network and will probably make source code public soon. The new app is a sort of cat and mouse simulation and may be more interesting than 4 blinking dots on a screen :)
Finally, I'd like to welcome Darwin Bruning from the Netherlands on our team. Darwin is studying biology and is helping a lot with understanding how biological neurons and our brain in general work.

12 Oct. 2002: Posted by Olivier Reisch (doctomoe@naria.karasuma.net)

A new version of the N.A.R.I.A. Library source code has been released. It contains a lot of bug fixes over the first version and includes new features as well, including the ability to stop learning once it has been trained successfully. There has also been a full API change to the library's functions to avoid conflicts with other libraries.

06 Oct. 2002: Posted by Olivier Reisch (doctomoe@naria.karasuma.net)

Success! The first version of the N.A.R.I.A. Library has been coded and it seems to be able to learn very simple tasks on its own, given simple "good" or "bad" feedback. The source code is available in the downloads section. Let's also welcome Joseph Vidal-Rosset as the project's philosoper. He's a philosophy professor at the University of Dijon in France and shares a particular interest for the philosophical questions around Artificial Intelligence.

27 Sep. 2002: Posted by Olivier Reisch (doctomoe@naria.karasuma.net)

This is not really the first day of the project, but it's the first official day of it's existence as I have started putting this web site up to document on the N.A.R.I.A. project.
Progress is pretty slow at the moment, I haven't started recruiting volunteers yet and next to that I am a busy person in general. So far I have started collecting documentation for the project and I also started drafting some basic ideas and blueprints for the NN. There's no source code yet.