Consider the logistic regression classifier applied to a sample x (bias feature added) with correct label y using weights vector w (bias corrected). The forward propagation for calculating the sample's loss is: z = wx a = o(z) L = log (a (1 a)) . Calculate the gradient of the loss using the chain rule Vw v (L) = Vw (2) V(a) V(L) and express it in terms of x, a, and y.