1
$\begingroup$

According to VAE paper : https://arxiv.org/pdf/1906.02691 (eq 2.10, page 21) VAE loss contain the expectation term. If I understand correctly, we need sample more than 1 result from an input x and average it's loss to satisfy the expectation term. But I find many implementation that just only sample one result, but why?

$\endgroup$
2
  • 2
    $\begingroup$ Though only using one sample will have a high variance, its expectation is still correct; this is a broader issue beyond VAEs and you may want to use search terms like "Monte Carlo Approximation", "Stochastic Gradient Descent" or "Finite Sum Problem". $\endgroup$ Commented Sep 22 at 18:55
  • $\begingroup$ @JohnMadden it's very helpful, thanks a lot. $\endgroup$
    – Manh
    Commented Sep 22 at 18:57

0

Your Answer

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

Browse other questions tagged or ask your own question.