[Re] Variational Sparse Coding

The reproduced paper proposes an improvement over the Variational Autoencoder (VAE) architecture by explicitly modelling sparsity in the latent space with a Spike and Slab prior distribution and drawing ideas from sparse coding theory. The main motivation behind the original work lies in the ability to infer truly sparse representations from generally intractable non-linear probabilistic models, simultaneously addressing the problem of lack of interpretability of latent features. Moreover, the proposed model improves the classification accuracy using the low-dimensional representations obtained, and significantly adds robustness while varying the dimensionality of latent space.

pdf