Note that:

The ReLU activation function is defined as follows:

For an input $x,$ReLU$(x)=$max$(0,x)$

The Softmax function is defined as follows:

Given the inputs, $x_i,i=1,$dots,$n$; the outputs are as follows:

$s(x_i)=\frac{e^{x_i}}{{\displaystyle \underset{j=1}{\overset{n}{}}}{e}^{x_j}}$

Suppose that the weights of the network are

$w11=-0\mathrm{.}5$; $w12=-1\mathrm{.}0,w21=1\mathrm{.}2$; $w22=1\mathrm{.}0$; $w31=-0\mathrm{.}6$; $w32=1\mathrm{.}0$

$h11=-0\mathrm{.}5$; $h12=-1\mathrm{.}0$; $h21=1\mathrm{.}0,h22=1\mathrm{.}5$; $h31=0\mathrm{.}4$; $h32=0\mathrm{.}3$

For an input of $x1=1\mathrm{.}0$ and $x2=1\mathrm{.}0$

i$.$ Calculate the net inputs a$1$ and a$2.$

ii$.$ Calculate the ReLU outputs.

iii. Calculate y$1$ and y$2.$

iv$.$ Calculate $z1$ and $z2.$