Question B$1[30$ marks$]$

$($a$)$ Consider a Multilayer Perceptron $($MAP$)$ network with $4$ input neurons, two hidden layers $(4$ neurons each$),2$ output neurons, ReLU hidden activation function and Softmax output activation function, draw the network archite and write an equation for computing the output $)=(f_{}(x)$ using matrix representations and $b^{(i)}$ as weights and biases matrices at layer $i).$

$($b$)$ Describe gradient descent and compare it to stochastic gradient descent and minibatch stochastic gradient descent.

$[10$ marks$]q,$

marks