I'm unsure how the softmax function comes about. I understand that the sigmoid function is a result of phrasing the log-odds as a linear equation and then rearranging the formula to model for the probability of an event occurring, but I'm not sure how to do this for softmax. Can someone point me in the right direction?

  • $\begingroup$ Are you dissatisfied with the explanation in en.wikipedia.org/wiki/Softmax_function ? $\endgroup$ – Andreas Nov 10 '17 at 18:13
  • $\begingroup$ @Andreas the answer wasn't really contained in there as it just states "this is a composition of K linear functions", however, the explanation I was looking for was presented here: en.wikipedia.org/w/… $\endgroup$ – tryingtolearn Nov 13 '17 at 9:59

Answer on how logistic regression is extended to the multinomial case is presented here:



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.