Semi-Implicit Graph Variational Auto-Encoders

TL;DR

SIG-VAE introduces a hierarchical semi-implicit variational framework for graph auto-encoders, enhancing modeling flexibility and interpretability, and significantly improving performance on graph tasks.

Contribution

It proposes a semi-implicit hierarchical variational auto-encoder for graphs, enabling better modeling of complex dependencies and implicit posteriors.

Findings

Outperforms state-of-the-art methods on graph tasks

Provides more interpretable latent representations

Captures complex graph properties effectively

Abstract

Semi-implicit graph variational auto-encoder (SIG-VAE) is proposed to expand the flexibility of variational graph auto-encoders (VGAE) to model graph data. SIG-VAE employs a hierarchical variational framework to enable neighboring node sharing for better generative modeling of graph dependency structure, together with a Bernoulli-Poisson link decoder. Not only does this hierarchical construction provide a more flexible generative graph model to better capture real-world graph properties, but also does SIG-VAE naturally lead to semi-implicit hierarchical variational inference that allows faithful modeling of implicit posteriors of given graph data, which may exhibit heavy tails, multiple modes, skewness, and rich dependency structures. Compared to VGAE, the derived graph latent representations by SIG-VAE are more interpretable, due to more expressive generative model and more faithful inference enabled by the flexible semi-implicit construction. Extensive experiments with a variety of graph data show that SIG-VAE significantly outperforms state-of-the-art methods on several different graph analytic tasks.