Variational Autoencoder - Reparameterization of the normal sampling

161 Views Asked by At

I was reading up on variational autoencoders and got confused at the part where they said the two expressions are equivalent.

$q(z|x)= N(\mu, \sigma^2) \\ q(z|x)= \mu + \sigma \circ N(0,1) $

Is there a proof anywhere that I can read about this ?

http://blog.fastforwardlabs.com/2016/08/22/under-the-hood-of-the-variational-autoencoder-in.html

enter image description here