Learning Autoencoders with Relational Regularization

Abstract

A new algorithmic framework is proposed for learning autoencoders of data distributions. We minimize the discrepancy between the model and target distributions, with a \emph{relational regularization} on the learnable latent prior. This regularization penalizes the fused Gromov-Wasserstein (FGW) distance between the latent prior and its corresponding posterior, allowing one to flexibly learn a structured prior distribution associated with the generative model. Moreover, it helps co-training of multiple autoencoders even if they have heterogeneous architectures and incomparable latent spaces. We implement the framework with two scalable algorithms, making it applicable for both probabilistic and deterministic autoencoders. Our relational regularized autoencoder (RAE) outperforms existing methods, $e.g.$, the variational autoencoder, Wasserstein autoencoder, and their variants, on generating images. Additionally, our relational co-training strategy for autoencoders achieves encouraging results in both synthesis and real-world multi-view learning tasks. The code is at https://github.com/HongtengXu/ Relational-AutoEncoders.

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Deep Unsupervised Clustering with Gaussian Mixture Variational Autoencoders
• Sliced-Wasserstein Autoencoder: An Embarrassingly Simple Generative Model
• Hierarchical Optimal Transport for Document Representation
• Scalable Gromov-Wasserstein Learning for Graph Partitioning and Matching
• Multi-view Generative Adversarial Networks
• Hierarchical Optimal Transport for Multimodal Distribution Alignment
• Fused Gromov-Wasserstein distance for structured objects: theoretical foundations and mathematical properties
• Gromov-Wasserstein Factorization Models for Graph Clustering

Cited by in corpus (7)

• Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety
• The Unbalanced Gromov Wasserstein Distance: Conic Formulation and Relaxation
• EfficientTDNN: Efficient Architecture Search for Speaker Recognition
• A contribution to Optimal Transport on incomparable spaces
• Personalised Federated Learning On Heterogeneous Feature Spaces
• Hierarchical Optimal Transport for Robust Multi-View Learning
• Making transport more robust and interpretable by moving data through a small number of anchor points