$1-)(25$ points$)$ In the following neural network, we assume that all node uses non$-$linear sigmoid activation functions.

u$)u$

Training data set :

$\backslash$table$[[,$input data$],[,i=1,i=2,i=3,i=4$