libdeep-python adds a Python API to the libdeep deep learning library, which is written in C. This allows you to obtain the raw processing speed of native C code while also having the convenience of being able to develop your application in Python. After a system has been trained you can export it as a standalone C commandline program which takes the input values as arguments and outputs the results to stdout.
To install dependencies on a Debian based system:
sudo apt-get install build-essential gnuplot doxygen python3-dev
Or on an Arch based system:
sudo pacman -S gcc gnuplot doxygen
Then install libdeep:
git clone https://github.com/bashrc/libdeep
cd libdeep
make
sudo make install
Finally install the python interface:
git clone https://github.com/bashrc/libdeep-python
cd libdeep-python
make
sudo make install
For example use cases see the examples directory and also the manpage.
man libdeep-python
To check that the system is working you can try the simplest example, which is learning the XOR function.
cd examples/xor
./xor.py