GPU Accelerated Closed-Loop Deep Learning

This is a GPU-accelerated version of the Closed-Loop Deep Learning library.

Multithreaded processing using a CUDA-enabled GPU allows for much more complex CLDL networks.

Prerequisites

A CUDA-enabled GPU is required to use this library.

The CUDA developer toolkit is required to compile and run the library.

Install instructions for Windows can be found here.

Install instructions for Linux can be found here.

Building

CLDL_CUDA uses cmake. Just enter the CLDL_CUDA directory from the root:

• cd CLDL_CUDA

and type:

• mkdir build && cd build
• cmake ..
• make

Test suite:

A gtest test suite is included in the build/gtest directory. The executable Google_Tests_run will be generated automatically when building CLDL_CUDA. Run the tests by doing:

• cd build/gtest
• ./Google_Tests_run

License

GNU GENERAL PUBLIC LICENSE

Version 3, 29 June 2007