Now consider back propagation to compute gradients. What is a CORRECT formula for computing @L

@$!12?$

A$.$ @L

@$!12=$ @L

@u$1+$ @L

@u$2$

B$.$ @L

@$!12=$ @L

@y$1$ @y$1$

@u$1$ @u$1$

@$!12$

C$.$ @L

@$!12=$ P$4$

k$=1$

@L

@y$$k $$ @y$$k

@u$2$ @u$2$

@$!12$

D$.$ @L

@$!12=$ P$4$

k$=1$

@L

@$1$k $$ @$1$k

@u$1$ @u$1$

@$!12$

For the next four problems, you will consider using Tensorflow to implement the

above neural network model. Suppose the training data is pre$-$processed and stored

as Xtrain $($predictor variables$)$ and ytrain $($labels$).$ All the needed packages and

functions have been imported, and are ready to use.