Why the exponent function is used in softmax?

#1

We could have directly used the value we got for calculating the probability. What is the significance of sing the exponent here?

#2

Hey @nikhil_sarda

I’m a little confused about your doubt. How can we calculate the probability without using the exponent function.
image
The exponent function is necessary for probability evaluation.
Request you to explain a bit more.

#3

if our output in last layer is x1,x2,x3. Then probability is x1/(x1+x2+x3), and so on.This will also ensure total prob as 1 and why exponential one is better? This is my doubt.

#4

What if one or more outputs among these are negative? Then you may think to just take the absolute value of the outputs, but, that to may prove to be problematic. e.g. Large negative values and large positive values both will correspond to a high probability (which is inherently problematic). There are other considerations too (mathematical ones), like, what will be the shape of the loss curve? Is it convex?
If you can solve these concerns, there is no harm in trying what you have suggested. You can compare the results and choose the better performing one.

1 Like
#5

I hope I’ve cleared your doubt. I ask you to please rate your experience here
Your feedback is very important. It helps us improve our platform and hence provide you
the learning experience you deserve.

On the off chance, you still have some questions or not find the answers satisfactory, you may reopen
the doubt.