Answer Happy

Question 20 20 pts a Suppose we have a neural network as shown below, with mean square error loss. This neural network is trained on SGD with one sample (x, y} where x=(1, 1), y=1. The SGD has a learning rate, which equal to 0.1. Compute the values of all weights (w1-w9) after performing an SGD update with current network settings. Input RELU Output W = 2 M W2 = 1 *1=1 W7 = 2 W3 = -2 Wa=3 22 Ypred W = W, = -1 X2=1 Ws = 1 Wo = 2 23 Note that • We assume the output layer has no activation function; . Using the chain rule to express the feedforward and backpropagation processes • Detailed processes are needed to gain full credits. . You can work this question with a word file, then upload it