The logistic or sigmoid function g(z) is given by 1/(1 + e). Using 2 (close enough for graphing) as an approximation for e (to make the math easy) compute g(z) for z = 0 ,1 ,2 ,4 JL , 72 ,-4 . Give the answers as simple fractions, for example, 3/4. Based on your 5 results, what alphabet letter shape is closest to a simple approximation to the function g(x)? The purpose of the sigmoid function is to introduce a between stages of the linear matrix transformations. Neural networks are trained by gradient descent. Gradient descent relies on back propagation to make the required differentiation computationally tractable. One of the nice properties of the sigmoid function is that it is so easy to