Please help me with task $1$ and $2!$

Problem $2$: Backpropagation

Consider a neural network with one input layer, one hidden layer, and one output layer. The input layer has two neurons, the hidden layer has two neurons $($plus a bias neuron$),$ and the output layer has one neuron. The activation function is the sigmoid function, $(x)=\frac{1}{1+e^{-x}}.$

$1$

Network Structure

Inputs: $x_1,x_2.$

Weights from input to hidden layer: $w_{11},w_{12},w_{21},w_{22}.$

Weights from hidden to output layer: $w_{31},w_{32}.$

Bias weights: $b_1,b_2$ for hidden layer, $b_3$ for output layer.

Desired output: $d.$

Loss function: $L=\frac{1}{2}(y-d{)}^2,$ where $y$ is the network output.

If you are more comfortable with matrix$-$vector multiplication, then the hidden layer translates $x=(x_1,x_2{)}^{TT}$ to $h=(Wx+b),$ where $W$ is a $22$ matrix with entries equal to the weights in the hidden layer, and $b=(b_1,b_2{)}^{TT}.$ The output layer outputs $d=(w_{31},w_{32})h+b_3.$

Tasks

Calculate the output of the network $(y)$ as a function of inputs, weights, and biases.

Derive the expressions for the gradients of the loss function with respect to each weight using backpropagation $(\frac{\text{d}\text{e}\text{l}\text{L}}{del{w}_{ij}}).$