Implementation of neural network from Andrew Ng as from Coursera
Author tomaskrehlik
5 Stars
Updated Last
3 Years Ago
Started In
October 2012

Implementation of label neural network

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.

What works

  • 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