I am working on this problem in which I have a dataset of $n$-dimensional examples that come from different and unknown distributions. Given a new sample, I wish to find $k$ examples from the dataset that come from distribution(s) closest to the new sample. Which measure (Kullback-Leibler vs Hellinger Distance) might be more suitable for this and why?
$\begingroup$
$\endgroup$
2
-
$\begingroup$ This could help: stats.stackexchange.com/questions/296361/… $\endgroup$– kjetil b halvorsen ♦Commented Aug 31, 2017 at 19:50
-
1$\begingroup$ Does this answer your question? Differences between Bhattacharyya distance and KL divergence $\endgroup$– Estimate the estimatorsCommented Sep 29, 2020 at 1:54
Add a comment
|