Q5 (20 points) Consider the following neural network that uses the natural log (i.e., ln) as the activation functions. Loss=squarelossAllweightsat0.33Input(X)11ln1 (10 points) Based on this one input and expected output, which of the weights is the Loss most sensitive to? Show a sorted list of weights based on their sensitivity to the overall Loss. Clearly show all calculations and your argument. Solution: (10 points) In your opinion, is natural log (i.e., In) a good activation function? Why or why not? Please explain clearly. Solution