libdfr-neural

neural network library
daniel ford 2015

libdfr-neural is a C++ library that implements fully connected neural networks with backpropagation for training

networks are composable with as many layers as desired

current layer types: linear, sigmoid, hyperbolic tangent, softmax

see the XOR test wrapper for example usage

the MNISTdigit wrapper shows how to use the library to recognize handwritten digits in the MNIST digit set

Name		Name	Last commit message	Last commit date
Latest commit History 33 Commits
.gitattributes		.gitattributes
.gitignore		.gitignore
DataLoader.cpp		DataLoader.cpp
DataLoader.h		DataLoader.h
README.md		README.md
dfrNeuralLayer.cpp		dfrNeuralLayer.cpp
dfrNeuralLayer.h		dfrNeuralLayer.h
dfrNeuralLinearLayer.cpp		dfrNeuralLinearLayer.cpp
dfrNeuralLinearLayer.h		dfrNeuralLinearLayer.h
dfrNeuralNet.cpp		dfrNeuralNet.cpp
dfrNeuralNet.h		dfrNeuralNet.h
dfrNeuralReLULayer.cpp		dfrNeuralReLULayer.cpp
dfrNeuralReLULayer.h		dfrNeuralReLULayer.h
dfrNeuralSigmoidLayer.cpp		dfrNeuralSigmoidLayer.cpp
dfrNeuralSigmoidLayer.h		dfrNeuralSigmoidLayer.h
dfrNeuralSoftmaxLayer.cpp		dfrNeuralSoftmaxLayer.cpp
dfrNeuralSoftmaxLayer.h		dfrNeuralSoftmaxLayer.h
dfrNeuralTanhLayer.cpp		dfrNeuralTanhLayer.cpp
dfrNeuralTanhLayer.h		dfrNeuralTanhLayer.h
dfrNeuralTestWrapper_MNISTdigit.cpp		dfrNeuralTestWrapper_MNISTdigit.cpp
dfrNeuralTestWrapper_XOR.cpp		dfrNeuralTestWrapper_XOR.cpp
train.txt		train.txt

krylenko/libdfr-neural