Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via Ranking

Abstract

Methods that learn representations of nodes in a graph play a critical role in network analysis since they enable many downstream learning tasks. We propose Graph2Gauss - an approach that can efficiently learn versatile node embeddings on large scale (attributed) graphs that show strong performance on tasks such as link prediction and node classification. Unlike most approaches that represent nodes as point vectors in a low-dimensional continuous space, we embed each node as a Gaussian distribution, allowing us to capture uncertainty about the representation. Furthermore, we propose an unsupervised method that handles inductive learning scenarios and is applicable to different types of graphs: plain/attributed, directed/undirected. By leveraging both the network structure and the associated node attributes, we are able to generalize to unseen nodes without additional training. To learn the embeddings we adopt a personalized ranking formulation w.r.t. the node distances that exploits the natural ordering of the nodes imposed by the network structure. Experiments on real world networks demonstrate the high performance of our approach, outperforming state-of-the-art network embedding methods on several different tasks. Additionally, we demonstrate the benefits of modeling uncertainty - by analyzing it we can estimate neighborhood diversity and detect the intrinsic latent dimensionality of a graph.

Updated: ICLR 2018 camera-ready version

References in corpus (8)

• Semi-Supervised Classification with Graph Convolutional Networks
• Inductive Representation Learning on Large Graphs
• Graph Embedding Techniques, Applications, and Performance: A Survey
• Deep Convolutional Networks on Graph-Structured Data
• Variational Graph Auto-Encoders
• Representation Learning on Graphs: Methods and Applications
• Word Representations via Gaussian Embedding
• A General Framework for Content-enhanced Network Representation Learning

Cited by in corpus (33)

• Scaling Graph Neural Networks with Approximate PageRank
• Is a Single Embedding Enough? Learning Node Representations that Capture Multiple Social Contexts
• Directed Graph Convolutional Network
• DDGK: Learning Graph Representations for Deep Divergence Graph Kernels
• A Comparative Study for Unsupervised Network Representation Learning
• LTE4G: Long-Tail Experts for Graph Neural Networks
• Indirect Adversarial Attacks via Poisoning Neighbors for Graph Convolutional Networks
• Robust Mid-Pass Filtering Graph Convolutional Networks
• Designing the Topology of Graph Neural Networks: A Novel Feature Fusion Perspective
• Pre-train and Learn: Preserve Global Information for Graph Neural Networks
• Relational Self-Supervised Learning on Graphs
• Geometer: Graph Few-Shot Class-Incremental Learning via Prototype Representation
• View-Invariant, Occlusion-Robust Probabilistic Embedding for Human Pose
• Learned Low Precision Graph Neural Networks
• A Framework for Generalizing Graph-based Representation Learning Methods
• COIN: Communication-Aware In-Memory Acceleration for Graph Convolutional Networks
• Reliable Node Similarity Matrix Guided Contrastive Graph Clustering
• NODE-SELECT: A Graph Neural Network Based On A Selective Propagation Technique
• Node Injection Attack Based on Label Propagation Against Graph Neural Network
• Semi-supervised Network Embedding with Differentiable Deep Quantisation
• EnGN: A High-Throughput and Energy-Efficient Accelerator for Large Graph Neural Networks
• SPINE: Structural Identity Preserved Inductive Network Embedding
• Clarify Confused Nodes via Separated Learning
• Topological based classification using graph convolutional networks
• Semi-Supervised Learning on Graphs Based on Local Label Distributions
• Bootstrap Equilibrium and Probabilistic Speaker Representation Learning for Self-supervised Speaker Verification
• Semantic Graph Neural Network with Multi-measure Learning for Semi-supervised Classification
• KMF: Knowledge-Aware Multi-Faceted Representation Learning for Zero-Shot Node Classification
• On the Impact of Communities on Semi-supervised Classification Using Graph Neural Networks
• Gaussian Embedding of Temporal Networks
• Topological Node2vec: Enhanced Graph Embedding via Persistent Homology
• CompanyKG: A Large-Scale Heterogeneous Graph for Company Similarity Quantification
• LSP : Acceleration and Regularization of Graph Neural Networks via Locality Sensitive Pruning of Graphs