1
$\begingroup$

Is it possible to train a simple perceptron with a threshold activation function such as this one: https://en.wikipedia.org/wiki/Perceptron with Backpropagation instead of the perceptron rule?

is it a problem that threshold functions are not defined around 0?

$\endgroup$
1
$\begingroup$

No, it is not possible. In order to use backprop you have to have non-zero derivatives of the function. Derivative of the step function is zero everywhere.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ The derivative of step function does not exist around the threshold value. $\endgroup$ – Osama El-Ghonimy Apr 26 at 13:06

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.