cppNet is small and easy to use library for neural networks. It is based on Eigen, C++ template library for linear algebra.
The library supports:
- Fully connected layers
- Convolution layers
- Max pooling layers
- Dropout
- SGD - stochastic gradient decent
- Loading and storing trained network
- Live graphs (Python)
Here are the steps to install cppNet.
- First thing you need are the repository files, get them by typing
git clone https://github.com/soCzech/cppNet.git
into terminal. - Run the
download.sh
script. It downloads MNIST dataset (used in the example files) and Eigen library which is needed for the vector operations. - You are ready to rock. You can try it by typing
make [all]
and runningfc
orcn
demo (see example usage in docs.pdf).
To find out how to build your own network, see test/*test.cpp
files and documentation in docs directory. For technical details, see the source files.