We want to approximate the following $($sigmoid$-$like$)$ function, given

in $($a$)$ in blue, using two ReLU functions $($b$)$ and $($c$)($in red$).($ReLU

formula: max$(w**x+b,0)$

$[10$ pts$]$ compute the parameters of ReLU function $($b$)$ and $($c$),$

respectively. Explain the way these ReLU functions approximate the

original function.

$[10$ pts$]$ show the weight and bias values in the following neural

network that implement above function. Show the output formula of

each node, and explain how the network implements the function.