changed:
-
----

=== Possible Mentors ===

 *Georg Holzmann
 *Ben Bogart


=== Description ===

 Pd has already an external collection called ann, which has support for feedforward networks
 and a basic self organizing map (SOM).
 The code of pds current neural network library can be found [http://pure-data.svn.sourceforge.net/viewvc/pure-data/trunk/externals/ann/ here].

There are two goals of this project:

 *extend the Pd bindings to the Fast Artificial Neural Network library ([http://leenissen.dk/fann/ FANN]) and include new FANN features like self organizing maps, growing neural gas and etc.

 *add support for recurrent neural networks (reservoir computing networks, ESN), which are very well suited for time series processing, and create Pd bindings to the [http://aureservoir.sourceforge.net/ aureservoir library]

Resources to start:

 *http://iem.kug.ac.at/pd/externals-HOWTO : a tutorial on how to write C externals for pure data

 *http://pure-data.svn.sourceforge.net/viewvc/pure-data/trunk/externals/ann :    the current artificial neural network code of pd

 *http://aureservoir.sourceforge.net : documentation of the aureservoir library

 *http://leenissen.dk/fann/html/files2/gettingstarted-txt.html : tutorial of the  FANN library

=== Required Skills ===

 *C/C++ programming

 *interest in artificial neural networks