Why did we use Z3 to compute cost? (Last assignment)


I just wondered why we used Z3 to compute cost without softmax regression process? Since we have learned that y" vector is needed to compute Loss function. How can we be sure that Z[L] is also same value with softmaxed Z[L]?

PS: We did not even use sigmoid function in the last activation step of the forward propagation function.


Hi, @berkaykepekci.

Here’s a great explanation by @paulinpaloalto.

Good luck with the course :slight_smile: