regression is a component of the FoxFire
project that handles regularized linear regression.
This module provides implementations of L1 and L0 regularized regression, with optional GPGPU acceleration via OpenCL.
This module currently supports Windows, OS X and Linux.
This module is header-only meaning
there is no installation step -- just clone the git repository and #include
the appropriate source files in your C++ projects. Just be sure to include / link against the
appropriate dependencies, as outlined in the Dependencies section below.
This module contains both generic and OpenCL specific components.
The base module requires the Eigen3
linear algebra library. Eigen3 is header-only, so the only additional step
required is the include the Eigen3 root directory eigen3 during compilation. For example,
on Linux systems one would pass the -I/usr/include/eigen3 to the appropiate compilier.
A sample GCC configuration might look like g++ $PROGRAM -I/usr/include/eigen3
In order to use GPGPU acceleration a valid OpenCL installation and at least one
OpenCL compatible compute device are required. In addition the Eigen3
and ViennaCL libraries need to installed. Some
special macros need to passed during program compilation, these are VIENNACL_WITH_OPENCL
and VIENNACL_WITH_OPENCL.
A sample GCC command might look like g++ $PROGRAM -DVIENNACL_WITH_OPENCL -DVIENNACL_WITH_EIGEN -I/usr/include/eigen3 -I/usr/include/viennacl -I/usr/local/cuda/include -L/usr/local/cuda/lib64 -lOpenCL
The FoxFire-regression package is licensed under the MIT license. To
view the MIT license please consult LICENSE.txt.
FoxFire-regression is based on the FOS
package produced by Johannes Lederer, Néhémy Lim and Saba Noorassa as part of the
HDIM research group.