This is (so far partial) a Julia implementation of the neural network that was originally written for MATLAB/Octave by Andrew Ng for Coursera Machine Learning Class. My implementation is hopefully more versatile for more than one layer.

• There is a nnCostFunction (see the sourcefile for documentation) which seems to be working, output is (cost, array of gradients)

• I was unable to use the optim.jl to make it iterate
• Add the checking gradient

• Just download and include load("neuralnetworks.jl") in your source