Milestone 4

  +150

Complete the back-propagation process for mazur-nn.cc, and use assertions to check the results against Mazur’s example.

To be clear, this means calculating new values for all eight weights. Here’s the correspondence between the variable names used in Mazur’s article and the variables in the C++ program:

  • hiddenOutputWts[0][0] is \(w_5\) (this one is calculated for you, the result is \(0.35891648\))
  • hiddenOutputWts[1][0] is \(w_6\)
  • hiddenOutputWts[0][1] is \(w_7\)
  • hiddenOutputWts[1][1] is \(w_8\)
  • inputHiddenWts[0][0] is \(w_1\)
  • inputHiddenWts[1][0] is \(w_2\)
  • inputHiddenWts[0][1] is \(w_3\)
  • inputHiddenWts[1][1] is \(w_4\)