The speedynn package is a Python library that aims at accelerating feature selection for nearest neighbor models via modern many-core devices such as graphics processing units (GPUs). The implementation is based on OpenCL.
The speedynn package is tested under Python 2.6 and Python 2.7. The required Python dependencies are:
- NumPy >= 1.6.1
Further, Swig, OpenCL, setuptools, and a working C/C++ compiler need to be available.
The package can be installed by executing the Makefile in the speedynn subdirectory, i.e.:
cd speedynn
make
Further, the package root needs to be added to the PYTHONPATH. On Debian/Ubuntu systems, this line should be added to .bashrc:
export PYTHONPATH=$PYTHONPATH:~/speedynn
in case the speedynn directory is located in the home folder. We recommend to use virtualenv to generate an independent environment for the package.
The source code is published under the GNU General Public License (GPLv2). The authors are not responsible for any implications that stem from the use of this software.