2
$\begingroup$

I am using a python parameter optimization library

https://keras-team.github.io/keras-tuner/documentation/hyperparameters/

And here were have the option to define a sampling distribution for various parameters. As i understand this essentially means that we will randomly draw a parameter from one end of the bounds more than the other end. However i'm not sure what this favors. For plain logarithmic sampling, does this mean that we will sample smaller values more often than larger values? Or the other way around?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

When using logarithmic sampling, it means that you are sampling uniformly on the log-transformed interval.

Here are 20 random values logarithmically sampled between 0 and 10:
enter image description here

And here are those same values plotted on the log-axis: enter image description here

So, quick answer, small values are sampled more frequently!

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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