Hi,
Comparing the explicit GMM VAE code with the loss equation from the blog, I'm confused about the missing term :
$\mathbb{E}_{q(y, z | x)} [\ln \frac{p(y)}{q(y | x)} ] $
(expectation of log(p(y)/q(y|x))-- the first term)
from the loss calculation here:
|
def labeled_loss(x, px_logit, z, zm, zv, zm_prior, zv_prior): |
|
loss = tf.add_n([nent] + [qy[:, i] * losses[i] for i in xrange(k)]) |
and in the notebook as well.
It would be great if you could clarify the reason for that
Thanks
Hi,
Comparing the explicit GMM VAE code with the loss equation from the blog, I'm confused about the missing term :
(expectation of log(p(y)/q(y|x))-- the first term)
from the loss calculation here:
vae-clustering/shared_subgraphs.py
Line 31 in 14525a9
vae-clustering/gmvae_k.py
Line 62 in 14525a9
and in the notebook as well.
It would be great if you could clarify the reason for that
Thanks