A fully disentangled variational auto-encoder (VAE) aims to identify disentangled latent components from observations. However, enforcing full independence between all latent components may be too strict for certain datasets. In some cases, multiple factors may be entangled together in a non-separable manner, or a single independent semantic meaning could be represented by multiple latent components within a higher-dimensional manifold. To address such scenarios with greater flexibility, we develop the Partially Disentangled VAE (PDisVAE), which generalizes the total correlation (TC) term in fully disentangled VAEs to a partial correlation (PC) term. This framework can handle group-wise independence and can naturally reduce to either the standard VAE or the fully disentangled VAE. Validation through three synthetic experiments demonstrates the correctness and practicality of PDisVAE. When applied to real-world datasets, PDisVAE discovers valuable information that is difficult to find using fully disentangled VAEs, implying its versatility and effectiveness.


    Access

    Download


    Export, share and cite



    Title :

    A Revisit of Total Correlation in Disentangled Variational Auto-Encoder with Partial Disentanglement


    Contributors:
    Li, Chengrui (author) / Wang, Yunmiao (author) / Wang, Yule (author) / Li, Weihan (author) / Jaeger, Dieter (author) / Wu, Anqi (author)

    Publication date :

    2025


    Type of media :

    Preprint


    Type of material :

    Electronic Resource


    Language :

    English




    Disentangled Variational Auto-Encoder for Semi-supervised Learning

    Li, Yang / Pan, Quan / Wang, Suhang et al. | ArXiv | 2017

    Free access

    AVAE: Adversarial Variational Auto Encoder

    Plumerault, Antoine / Borgne, Hervé Le / Hudelot, Céline | ArXiv | 2020

    Free access

    Variational auto encoder for mixed data types

    European Patent Office | 2020

    Free access

    Disentangling Latent Factors of Variational Auto-Encoder with Whitening

    Hahn, Sangchul / Choi, Heeyoul | ArXiv | 2018

    Free access