Dual Graph Representation Learning
This addresses the limitation of existing graph embedding methods that are transductive or neglect node contexts, potentially improving downstream applications in graph analysis.
The paper tackles the problem of graph representation learning for unseen nodes and across different graphs by introducing CADE, a context-aware unsupervised dual encoding framework that combines real-time neighborhoods with neighbor-attentioned representation and preserves memory of known nodes, showing effectiveness compared to state-of-the-art methods.
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.