Prediction

The figure below represents an architecture of artificial neural network with sigmoid activation function $f(s)=\frac{1}{1+e^{-s}}$

The training data used for this architecture is:

$\backslash$table$[[$X$1,$X$2,$X$3,$Desired Output$],[0\mathrm{.}1,0\mathrm{.}4,0\mathrm{.}7,0\mathrm{.}02]]$

The initial weights are:

$\backslash$table$[[$W$1,$W$2,$W$3,$b$],[0\mathrm{.}4,0\mathrm{.}3,0\mathrm{.}6,2\mathrm{.}3]]$

a$-$ Calculate the updated weights $W_i$ using the retropropagation algorithm $(20$pts$)$

b$-$ what are the needed retroprapagation algorithm$(2$ pts$)$