GLSR-VAE: Geodesic Latent Space Regularization for Variational AutoEncoder Architectures

TL;DR

This paper introduces GLSR-VAE, a regularization technique for VAEs that enables continuous control over data attributes in the latent space, demonstrated through a music generation task.

Contribution

The paper presents a novel geodesic latent space regularization method for VAEs, allowing fine control of data attributes during generation.

Findings

Regularization improves attribute control in generated data.

Enables continuous modulation of data attributes.

Effective in a music sequence variation task.

Abstract

VAEs (Variational AutoEncoders) have proved to be powerful in the context of density modeling and have been used in a variety of contexts for creative purposes. In many settings, the data we model possesses continuous attributes that we would like to take into account at generation time. We propose in this paper GLSR-VAE, a Geodesic Latent Space Regularization for the Variational AutoEncoder architecture and its generalizations which allows a fine control on the embedding of the data into the latent space. When augmenting the VAE loss with this regularization, changes in the learned latent space reflects changes of the attributes of the data. This deeper understanding of the VAE latent space structure offers the possibility to modulate the attributes of the generated data in a continuous way. We demonstrate its efficiency on a monophonic music generation task where we manage to generate variations of discrete sequences in an intended and playful way.