DLS 2 W3 Fitting Batch Norm to NN

In Implementing Gradient Descent (with Batch Norm)

  • about 10:23 in video

Should the description include -

  • for each mini-batch
  • initialize (before forward prop)
  • compute cost (before backward prop)


Hi dds,

You would start with initialization. Then for each minibatch you would perform forward prop, compute cost, then perform backward prop.