In the Error back-prop training alg 2" slide 11 of lecture 3 "multilayer perceptrons if all of the inputs h" to a particular output neuron y for a particular training example Xn were very small (less than 0.5), how would the raw value z of that neuron change if the second hidden layer activation function was changed to a ReLU rather than a sigmoid? Choose the most correct answer. Select one: Select one: O a. It would increase O b. It would decrease O c. It would go to zero O d. It would become infinitely large O e. It's impossible to say Of. all of the above O g. band c are correct Oh. a, b and care correct Oi. a and e are correct Clear my choice Check