3. Backpropagation. Consider the model with an M-dimensional input x=(x1,,xM) and scalar output y given by zky=j=1MAkjxj+bk,uk==1Kezezk,k=1,,K=k=1Kj=1MCkjxjuk+k=1Kdkuk with parameters Akj,bk,Ckj and dk. (a) You are given training samples xi=(xi1,,xiM) and yi. Write the equations relating the inputs xij to the output yi. (b) Draw the computation graph between the data, parameters and loss function using the loss function, J=i=1N(yiyi)2. (c) Show how to compute the components of J/z from the components of J/u