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 here.

There are two goals of this project:

• extend the Pd bindings to the Fast Artificial Neural Network library ( 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 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