Dart Deep Neural Network
git clone /~https://github.com/NotsOverflow/dartnet.git && cd dartnet
# generate a Xor dataset of 2500 data ( in, in, out )
dart .\tools\generate_xor.dart small.bin 2500
# train net [2,4,1] with small.bin dataset and save weights
dart .\main.dart train small.bin weight.bin 2 4
# use it!
dart .\main.dart run weight.bin small.bin out.bin 2 4 1
double learningRate = 0.15;
double step = 0.5;
double myActivation(double x){ ... }
double myDerivate(double x){ ... }
Net neuralNet = new Net(
[3,6,8,1],
activationFunction: myActivation,
derivativeFunction: myDerivate,
eta: learningRate,
alpha: step
); import 'libs/net.dart';
Net neuralNet = new Net([3,6,8,1]);
neuralNet.loadWeights("weightFile");
neuralNet.feedForward([0.2, 0.8, 1.0]);
List<double> result = neuralNet.getResults();