# A Degeneracy Framework for Scalable Graph Autoencoders

## TL;DR

This paper introduces a scalable framework for graph autoencoders and variational autoencoders that uses graph degeneracy to train on dense node subsets, significantly improving scalability and speed while maintaining performance.

## Contribution

It presents a novel degeneracy-based approach for scaling graph autoencoders and VAEs to large graphs, enabling training on millions of nodes efficiently.

## Key findings

- Achieves competitive results on large-scale graphs with up to millions of nodes.
- Improves training speed and scalability without sacrificing performance.
- First application of scalable graph AE and VAE models to very large graphs.

## Abstract

In this paper, we present a general framework to scale graph autoencoders (AE) and graph variational autoencoders (VAE). This framework leverages graph degeneracy concepts to train models only from a dense subset of nodes instead of using the entire graph. Together with a simple yet effective propagation mechanism, our approach significantly improves scalability and training speed while preserving performance. We evaluate and discuss our method on several variants of existing graph AE and VAE, providing the first application of these models to large graphs with up to millions of nodes and edges. We achieve empirically competitive results w.r.t. several popular scalable node embedding methods, which emphasizes the relevance of pursuing further research towards more scalable graph AE and VAE.