Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
148 views14 pages

Variational Autoencoders

Variational autoencoders (VAEs) are a type of autoencoder that learns latent variable probabilistic models. VAEs are trained to maximize a lower bound on the likelihood of the data, resulting in a latent space that has good properties for generation. The encoder outputs the parameters of a probability distribution rather than a single point, allowing the model to describe a range of possible latent representations. VAEs provide a probabilistic approach to describing observations in latent space and allow for generation of new examples based on the learned latent space.

Uploaded by

bhavana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
148 views14 pages

Variational Autoencoders

Variational autoencoders (VAEs) are a type of autoencoder that learns latent variable probabilistic models. VAEs are trained to maximize a lower bound on the likelihood of the data, resulting in a latent space that has good properties for generation. The encoder outputs the parameters of a probability distribution rather than a single point, allowing the model to describe a range of possible latent representations. VAEs provide a probabilistic approach to describing observations in latent space and allow for generation of new examples based on the learned latent space.

Uploaded by

bhavana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

Variational Autoencoders

By P Jayasri
Roll no: PC2113003013019
Definition of Variational Autoencoder

● A variational autoencoder (VAE) provides a probabilistic


manner for describing an observation in latent space.

● a variational autoencoder can be defined as being an


autoencoder whose training is regularised to avoid overfitting
and ensure that the latent space has good properties that
enable generative process.
continues..

Just as like standard autoencoder, a variational autoencoder is an


architecture composed of both an encoder and a decoder and that
is trained to minimise the reconstruction error between the
encoded-decoded data and the initial data.

rather than building an encoder which outputs a single value to


describe each latent state attribute, we'll formulate our encoder to
describe a probability distribution for each latent attribute.
Difference between autoencoder and variational autoencoder
Example
continues..
continues..
continues..
Statistical motivation

Suppose that there exists some hidden variable z which generates an


observation x.
now probabilistic versions of these two objects. The
“probabilistic decoder” is naturally defined by p(x|z), that
describes the distribution of the decoded variable given the
encoded one, whereas the “probabilistic encoder” is defined
by p(z|x), that describes the distribution of the encoded
variable given the decoded one.
we can also remind the the well-known Bayes theorem that makes the link between the

prior p(z), the likelihood p(x|z), and the posterior p(z|x)

require the use of approximation techniques such as variational inference.

for a given input x, we want to maximise the probability to have x̂ = x when we sample

z from the distribution q*_x(z) and then sample x̂ from the distribution p(x|z).
Variational autoencoder representation
Applications of autoencoder

1. Dimensionality Reduction
2. Image Compression
3. Image Denoising
4. Feature Extraction
5. Image generation
6. Sequence to sequence prediction
7. Recommendation system
Question bank

2 marks questions

1. Define variational auto encoder?


2. Differentiate autoencoder with variational autoencoders?
3. List applications of autoencoders

16 marks questions

1. Explain in detail about variational autoencoders?

You might also like