L-VAE: Variational Auto-Encoder with Learnable Beta for Disentangled Representation

TL;DR

L-VAE introduces a learnable hyperparameter mechanism within the VAE framework to optimize the balance between disentanglement and reconstruction, improving interpretability of latent representations.

Contribution

The paper presents L-VAE, a novel extension of {eta}-VAE that learns loss term weights and model parameters simultaneously for better disentanglement control.

Findings

L-VAE outperforms existing models on multiple datasets.

It achieves a better balance between reconstruction and disentanglement.

Qualitative results show effective disentangling of facial attributes.

Abstract

In this paper, we propose a novel model called Learnable VAE (L-VAE), which learns a disentangled representation together with the hyperparameters of the cost function. L-VAE can be considered as an extension of \b{eta}-VAE, wherein the hyperparameter, \b{eta}, is empirically adjusted. L-VAE mitigates the limitations of \b{eta}-VAE by learning the relative weights of the terms in the loss function to control the dynamic trade-off between disentanglement and reconstruction losses. In the proposed model, the weight of the loss terms and the parameters of the model architecture are learned concurrently. An additional regularization term is added to the loss function to prevent bias towards either reconstruction or disentanglement losses. Experimental analyses show that the proposed L-VAE finds an effective balance between reconstruction fidelity and disentangling the latent dimensions. Comparisons of the proposed L-VAE against \b{eta}-VAE, VAE, ControlVAE, DynamicVAE, and {\sigma}-VAE on datasets, such as dSprites, MPI3D-complex, Falcor3D, and Isaac3D reveals that L-VAE consistently provides the best or the second best performances measured by a set of disentanglement metrics. Moreover, qualitative experiments on CelebA dataset, confirm the success of the L-VAE model for disentangling the facial attributes.