Variational Autoencoder for MNIST

Test Task

Build a variational autoencoder for the MNIST data set. Compute the mean value and the standard deviation for all ten MNIST classes in your latent representation. Plot your latent representation, either directly or using PCA.

Creating the environment and running the code

Create a virtual environment using the package-list.txt Run the init_challenge.py

Results

The sample output of latent space representation and latent space representation along with original test points are stored in the sample_outputs folder.

Example: Clustering of digits in the latent space