Adversarial Disentanglement with Grouped Observations

2020/01/14 by József Németh, Nemeth, Jozsef
Computer Science · #Advanced Image Processing Techniques #Digital Media Forensic Detection #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · pdf · doi:10.48550/arxiv.2001.04761

openalex publication_date 2020/01/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We consider the disentanglement of the representations of the relevant attributes of the data (content) from all other factors of variations (style) using Variational Autoencoders. Some recent works addressed this problem by utilizing grouped observations, where the content attributes are assumed to be common within each group, while there is no any supervised information on the style factors. In many cases, however, these methods fail to prevent the models from using the style variables to encode content related features as well. This work supplements these algorithms with a method that eliminates the content information in the style representations. For that purpose the training objective is augmented to minimize an appropriately defined mutual information term in an adversarial way. Experimental results and comparisons on image datasets show that the resulting method can efficiently separate the content and style related attributes and generalizes to unseen data.

Citations

• Deep Convolutional Inverse Graphics Network
• Auto-Encoding Variational Bayes
• Semi-Supervised Learning with Deep Generative Models
• Gradient-based learning applied to document recognition
• Representation Learning: A Review and New Perspectives
• Weakly-supervised Disentangling with Recurrent Transformations for 3D View Synthesis
• Understanding disentangling in β-VAE
• Unsupervised Adversarial Invariance
• Learning Latent Subspaces in Variational Autoencoders
• Invariant Representations without Adversarial Training
• Recent Advances in Autoencoder-Based Representation Learning
• Disentangling Style and Content in Anime Illustrations
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

Related

• Semi-supervised Disentanglement with Independent Vector Variational Autoencoders
• Disentangling factors of variation in deep representations using adversarial training
• JADE: Joint Autoencoders for Dis-Entanglement
• Self-Supervised Disentanglement by Leveraging Structure in Data Augmentations
• Multi-Level Variational Autoencoder: Learning Disentangled\n Representations from Grouped Observations