This is an implementaion of neural networks written for educational purposes. It includes MLP(multi-layer perceptron), AE(auto-encoder), and RBM(restricted Boltzmann machine). It also contains CUDA code for forward propagation. But it's less efficient than CPU implementation because it was not optimized.
Prof. Injung Kim (http://deeplearning.handong.edu), School of CSEE, Handong Global University (http://www.handong.edu).
Donwload: git clone https://github.com/callee2006/HGUNeuralNetworks
Compile:
make // using CPU
make CUDA=ENABLE // enable GPU (less efficient than CPU)