Consider the following diagram with a input$-$output sample:

Inputs $\backslash (\backslash$rightarrow $\backslash )$ current drawn by an electrical device $($A$)$

$\backslash (\backslash$rightarrow $\backslash )$ temperature of the device $\backslash (\backslash$left$(\{\}^\{\backslash$circ$\}\backslash$mathrm$\{$C$\}\backslash$right$)\backslash )$

Output $\backslash (\backslash$rightarrow $\backslash )$ whether a fault has occurred or not

a$.$ To solve the above problem, propose a simple Deep MLP network with at least $5$ model parameters.

Mention the number of perceptrons, hidden layers, model parameters of your proposed network. Also, mention the activation functions of your network.

b$.$ Why activation functions are used in Neural Networks? Explain briefly mentioning some of the activation functions.

c$.$ For the given sample inputs, perform the forward propagation. $[$Initialize the model parameters with non$-$zero numbers between $0$ to $1]$

d$.$ For the above problem, choose a suitable loss function. Perform backward propagation to update the model parameters of one perceptron of the last layer.