GAGE: Geometry Preserving Attributed Graph Embeddings

TL;DR

This paper introduces GAGE, a novel tensor factorization method for attributed graph embeddings that preserves both network topology and node attribute geometry, leading to improved performance in downstream tasks.

Contribution

The paper presents a new tensor factorization approach that simultaneously preserves the geometry of connectivity and attributes in attributed network embeddings, which was previously challenging.

Findings

Significant performance improvements over state-of-the-art baselines.

Effective preservation of both network structure and attribute proximity.

Lightweight algorithm suitable for large attributed networks.

Abstract

Node embedding is the task of extracting concise and informative representations of certain entities that are connected in a network. Various real-world networks include information about both node connectivity and certain node attributes, in the form of features or time-series data. Modern representation learning techniques employ both the connectivity and attribute information of the nodes to produce embeddings in an unsupervised manner. In this context, deriving embeddings that preserve the geometry of the network and the attribute vectors would be highly desirable, as they would reflect both the topological neighborhood structure and proximity in feature space. While this is fairly straightforward to maintain when only observing the connectivity or attribute information of the network, preserving the geometry of both types of information is challenging. A novel tensor factorization approach for node embedding in attributed networks is proposed in this paper, that preserves the distances of both the connections and the attributes. Furthermore, an effective and lightweight algorithm is developed to tackle the learning task and judicious experiments with multiple state-of-the-art baselines suggest that the proposed algorithm offers significant performance improvements in downstream tasks.