Dual Graph Representation Learning

TL;DR

This paper introduces CADE, a dual encoding framework for graph representation learning that combines real-time neighborhood information with memory of known nodes, enabling better generalization and context-awareness.

Contribution

CADE is a novel framework that integrates real-time neighborhood and memory-based representations for more effective inductive graph embedding.

Findings

CADE outperforms state-of-the-art methods in experiments.

The dual encoding approach improves generalization to unseen nodes.

CADE effectively captures node context and memory for better embeddings.

Abstract

Graph representation learning embeds nodes in large graphs as low-dimensional vectors and is of great benefit to many downstream applications. Most embedding frameworks, however, are inherently transductive and unable to generalize to unseen nodes or learn representations across different graphs. Although inductive approaches can generalize to unseen nodes, they neglect different contexts of nodes and cannot learn node embeddings dually. In this paper, we present a context-aware unsupervised dual encoding framework, \textbf{CADE}, to generate representations of nodes by combining real-time neighborhoods with neighbor-attentioned representation, and preserving extra memory of known nodes. We exhibit that our approach is effective by comparing to state-of-the-art methods.