Sage-Tips

What is the difference between the variational Autoencoder and a regular Autoencoder?

An autoencoder accepts input, compresses it, and then recreates the original input. A variational autoencoder assumes that the source data has some sort of underlying probability distribution (such as Gaussian) and then attempts to find the parameters of the distribution.

Is Autoencoder a generative model?

Autoencoders on a high level are composed of an encoder, a latent space, and a decoder. An autoencoder is trained by using a common objective function that measures the distance between the reproduced and original data. Autoencoders have many applications and can also be used as a generative model.

Is variational autoencoder a generative model?

VAE’s, shorthand for Variational Auto-Encoders are class of deep generative networks which has the encoder (inference) and decoder (generative) parts similar to the classic auto-encoder.

What is Variational autoencoder (VAE)?

A variational autoencoder (VAE) is a type of neural network that learns to reproduce its input, and also map data to latent space. A VAE can generate samples by first sampling from the latent space. We will go into much more detail about what that actually means for the remainder of the article.

Do I need to learn more about variational inference to understand autoencoders?

While it is recommended to learn more about variational inference, it is not actually required to understand the implementation of variational autoencoders. To summarize, variational autoencoders combine autoencoders with variational inference. Let’s now look at the architecture of variational autoencoders.

What are generative adversarial networks (GANs)?

In a pr e vious post, published in January of this year, we discussed in depth Generative Adversarial Networks (GANs) and showed, in particular, how adversarial training can oppose two networks, a generator and a discriminator, to push both of them to improve iteration after iteration.

Why do autoencoders have to be adjusted for dimensionality?

For these two reasons, the dimension of the latent space and the “depth” of autoencoders (that define degree and quality of compression) have to be carefully controlled and adjusted depending on the final purpose of the dimensionality reduction. When reducing dimensionality, we want to keep the main structure there exists among the data.