Trjtdtk

What word means to make something obsolete?

Pulling the rope with one hand is as heavy as with two hands?

Do I have to worry about players making “bad” choices on level up?

Has any spacecraft ever had the ability to directly communicate with civilian air traffic control?

How does a Swashbuckler rogue "fight with two weapons while safely darting away"?

Modify locally tikzset

Help, my Death Star suffers from Kessler syndrome!

Colliding particles and Activation energy

Confusion about capacitors

Why does nature favour the Laplacian?

Can fracking help reduce CO2?

How to back up a running remote server?

Is it cheaper to drop cargo drop than to land it?

What is the strongest case that can be made in favour of the UK regaining some control over fishing policy after Brexit?

A question regarding using the definite article

A non-technological, repeating, visible object in the sky, holding its position in the sky for hours

What does YCWCYODFTRFDTY mean?

Find the coordinate of two line segments that are perpendicular

What does "rf" mean in "rfkill"?

Can a creature tell when it has been affected by a Divination wizard's Portent?

Was it really necessary for the Lunar Module to have 2 stages?

Phrase for the opposite of "foolproof"

How can Republicans who favour free markets, consistently express anger when they don't like the outcome of that choice?

Single Colour Mastermind Problem

How is the standard deviation of VAE's obtained?

How to predict the probability of an event?Behavioral Differences between Standard Autoencoder and Variational AutoencoderHow are the positions of the output nodes determined in the Kohonen - Self Organizing Maps algorithm?How do i use the Gaussian function with a Naive Bayes Classifier?Right Way to Input Text Data in Keras Auto EncoderAre the raw probabilities obtained from XGBoost, representative of the true underlying probabilties?what's the difference between autoencoder and autoassociative neural networks?How to convert an array of numbers into probability values?Help solving Bigram Model with the following probabilitiesHow to use Adaptive Rejection Sampling to Update Alpha in iGMM

0

$begingroup$

I am trying to build a Variational Autoencoder. I was looking at various codes online and found most of them in some way or another copy Francois Chollet (Google researchers) code.

Now my main question with this code is this part:

As you can clearly see the $log(sigma)$ is the output from a Dense layer. Where did this assumption come from (the output is $log(sigma)$ and not $sigma$? How is it possible that we generate a random normal distribution with standard deviation $sigma$ like this? Is it due to the way the computer generates Normal Distributions?

probability autoencoder gaussian

share|improve this question

edited Apr 9 at 9:28

DuttaA

asked Apr 8 at 12:09

DuttaADuttaA

511319

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @Esmailian you can see that K_log_sigma = dense(latent_dim) which is the same as z_mean taken in the above line
  $endgroup$
  – DuttaA
  Apr 8 at 12:28
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @Esmailian ok sorry..i'll edit that part out but the question will now be why is it $log(sigma)$ and not $sigma$
  $endgroup$
  – DuttaA
  Apr 8 at 12:37
  
  

  
  
  
  

add a comment | 

0

$begingroup$

I am trying to build a Variational Autoencoder. I was looking at various codes online and found most of them in some way or another copy Francois Chollet (Google researchers) code.

Now my main question with this code is this part:

As you can clearly see the $log(sigma)$ is the output from a Dense layer. Where did this assumption come from (the output is $log(sigma)$ and not $sigma$? How is it possible that we generate a random normal distribution with standard deviation $sigma$ like this? Is it due to the way the computer generates Normal Distributions?

probability autoencoder gaussian

share|improve this question

edited Apr 9 at 9:28

DuttaA

asked Apr 8 at 12:09

DuttaADuttaA

511319

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @Esmailian you can see that K_log_sigma = dense(latent_dim) which is the same as z_mean taken in the above line
  $endgroup$
  – DuttaA
  Apr 8 at 12:28
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @Esmailian ok sorry..i'll edit that part out but the question will now be why is it $log(sigma)$ and not $sigma$
  $endgroup$
  – DuttaA
  Apr 8 at 12:37
  
  

  
  
  
  

add a comment | 

$begingroup$

I am trying to build a Variational Autoencoder. I was looking at various codes online and found most of them in some way or another copy Francois Chollet (Google researchers) code.

Now my main question with this code is this part:

As you can clearly see the $log(sigma)$ is the output from a Dense layer. Where did this assumption come from (the output is $log(sigma)$ and not $sigma$? How is it possible that we generate a random normal distribution with standard deviation $sigma$ like this? Is it due to the way the computer generates Normal Distributions?

probability autoencoder gaussian

share|improve this question

edited Apr 9 at 9:28

DuttaA

asked Apr 8 at 12:09

DuttaADuttaA

511319

$endgroup$

share|improve this question

edited Apr 9 at 9:28

DuttaA

asked Apr 8 at 12:09

DuttaADuttaA

511319

share|improve this question

edited Apr 9 at 9:28

DuttaA

asked Apr 8 at 12:09

DuttaADuttaA

511319

asked Apr 8 at 12:09

DuttaADuttaA

511319

asked Apr 8 at 12:09

DuttaADuttaA

511319