Consider a multi-class logistic regression with two classes using the softmax function. Let zo = wx, and z₁ = wfx, and z

[answerhappygod]

Consider A Multi Class Logistic Regression With Two Classes Using The Softmax Function Let Zo Wx And Z Wfx And Z 1 (126.72 KiB) Viewed 8 times

Consider a multi-class logistic regression with two classes using the softmax function. Let zo = wx, and z₁ = wfx, and z = = [zo, z1]. Express the softmax outputs P(y = 0|x) = 90(20, 2₁) and P(y = 1|x) = 91 (20, 21) in terms of zo and 21. Note: Click on the ? symbol next to the input space to see what variables are allowed in your submission. go(20, 21) = 91 (20, 21): = symbolic expression symbolic expression Now, you will show that using two softmax outputs is equivalent to using one sigmoid output. Rewrite the expression for P(y = 1|x) in the form of a sigmoid, where the argument is in terms of zo and 2₁, i.e. P(y = 1|x) but z is expressed in terms of zo and 2₁. P(y = 1|x) : symbolic expression - The expression go (zo, 21) is equivalent to o(2), when ? z = symbolic expression