All Questions
5 questions
1
vote
0
answers
389
views
Classical VAE not learning 2D gaussian mixture distribution using MSE loss
I've been exploring VAE for non-image data. I consider small to medium-sized continuous vector spaces and I want to learn the distribution of a dataset in that space.
As a warm up exercise, I tried ...
- 113
1
vote
1
answer
228
views
Keras--variational auto-encoder in R studio, which part is defined as Encoder?
This is the example given on VAE, the circle part is something I do not understand. It defined the encoder part as from (X to Z_mean), but my understanding is from(x to Z). Or it just simply does not ...
- 445
27
votes
3
answers
26k
views
Loss function autoencoder vs variational-autoencoder or MSE-loss vs binary-cross-entropy-loss
When having real valued entries (e.g. floats between 0 and 1 as normalized representation for greyscale values from 0 to 256) in our label vector, I always thought that we use MSE(R2-loss) if we want ...
57
votes
5
answers
44k
views
How to weight KLD loss vs reconstruction loss in variational auto-encoder?
in nearly all code examples I've seen of a VAE, the loss functions are defined as follows (this is tensorflow code, but I've seen similar for theano, torch etc. It's also for a convnet, but that's ...
- 1,019
2
votes
1
answer
2k
views
Reparameterization trick for other non-normal distributions? (Example tensorflow code?) [closed]
Can anyone point me toward tensorflow code for the reparameterization trick for non-normal distributions? (uniform, beta, poisson, negative-binomial, etc?) Both for the sampling stage and the KL loss ...
