Question 4. (15 pts.) The following neural network has 4 layers (i.e., 1 input, 2 hidden, 1 output) with 9 units labeled 1,2,,9 where the units 1 and 2 are the input units (i.e., corresponding to input features) and the units 8 and 9 are the output units. W68,W69,W36,W78,W79,T3,T4,T5,T6,T7,T8,T9 following the notation in the slides. All units (of course, except the input units) use negative sigmoid functions, meaning that their output is computed using the equation, which differs from the standard sigmoid by a minus sign oj=1+exp((netjTj))1 whose derivative is given by netjoj=oj(1+oj) Derive the derivative w46 step-by-step using the backpropagation rule. Answer: w46Ed=