2017年2月6日 · We introduce beta-VAE, a new state-of-the-art framework for automated discovery of interpretable factorised latent representations from raw image data in a completely unsupervised manner.

OpenReview is a platform for hosting and managing academic venues, such as conferences and journals.

2020年9月28日 · The paper \textit {removes the inherent trade-off} between reconstruction accuracy and disentanglement for $\beta$-VAE. Existing methods, such as $\beta$-VAE and FactorVAE, assign a large weight to the KL-divergence term in the objective function, leading to high reconstruction errors for the sake of better disentanglement.

2018年9月27日 · However, we show that methods like beta-VAE simplify the tendency of variational inference to underfit causing pathological over-pruning and over-orthogonalization of learned components.

Specifically, we use diffusion-based autoencoder (Diff-AE) as a backbone while resorting to $\beta$-VAE as a co-pilot to extract semantically disentangled representations. The strong generation ability of diffusion model and the good disentanglement ability of VAE model are complementary.

2025年1月20日 · Several variants of Variational Autoencoders have been developed to address inherent limitations. Specifically, $\sigma$-VAE utilizes a scaled identity matrix $\sigma^2 I$ in the decoder variance, while $\beta$-VAE introduces a hyperparameter $\beta$ to reweight the negative ELBO loss. However, a unified theoretical and practical understanding of model optimality remains unclear. For example ...

2022年12月5日 · On the other hand, standard Variational Autoencoders (VAEs) typically have access to a low-dimensional latent space but exhibit poor sample quality. We present DiffuseVAE, a novel generative framework that integrates VAE within a diffusion model framework, and leverage this to design novel conditional parameterizations for diffusion models.

2022年1月28日 · As an important problem in causal inference, we discuss the identification and estimation of treatment effects (TEs) under limited overlap; that is, when subjects with certain features belong to a...

In the factor-vae setting, we expect that a very low beta places more emphasis on reconstruction quality and result in a high discriminator performance; since less e ort is spent on disentanglement.

2024年1月16日 · For alpha = 0, the alpha-TCVAE reduces to beta-VAE. Experiments show that alpha-TCVAE brings some improvement in disentanglement quality measures and diversity of generated samples. Comparisons with beta-TCVAE, beta-TCVAE, FactorVAE, beta-VAE+HFS, VAE and StyleGAN are provided.