An implementation of a feedforward neural network. Eigen is used for linear algrebra calculations. Tests are implemented using Catch.
The example in /examples trains a neural network to recognize images of handwritten digits, and reaches an accuracy of around 93%. The training data is from the MNIST data base of handwritten digits, which can be found here.