Consider the three$-$layer backpropagation network for solving the XOR operation below with initial weights and threshold levels set randorty as follows:

$w_{13}=0\mathrm{.}5,w_{14}=0\mathrm{.}9,w_{23}=0\mathrm{.}4,w_{24}=1\mathrm{.}0,w_{35}=-1\mathrm{.}2,w_{45}=1\mathrm{.}1,{}_3=0\mathrm{.}8,{}_4=-0\mathrm{.}1$ and ${}_5=$

$0\mathrm{.}3.$ Inputs are set to $x_1=1$ and $x_2=1.$ Desired output $=0.$ Activation function $=Y^{tanh}=\frac{2a}{1+e^{-bx}}-a,$ where $a=1\mathrm{.}716$ and $b=0\mathrm{.}667$ are constants.

a Calculate the actual outputs $y_3$ and $y_4$ of neurons $3$ and $4$ in the hidden layers.

b Calculate the actual output $y_5$ of neuron $5.$

c Calculate the error, e$,$ between the actual output and the desired output.

d Calculate the error gradient for neuron $5$ in the output layer.

e Determine the weight corrections for $w_{35},w_{45}$ and ${}_5,$ assuming that the learning rate parameter, $,$ is equal to $0\mathrm{.}1.$

f Calculate the new weight values for $w_{35},w_{45}$ and ${}_5.$