Neural Network Graph Similarity Computation Based on Graph Fusion

TL;DR

This work tackles the scalability challenge of graph-graph similarity by introducing Graph Fusion, which merges two graphs into a single large graph and applies a global attention mechanism (Transformer or Performer) to enable joint cross-graph interaction with linear-time variants. It then derives graph-level and node-level similarity vectors and fuses them through an end-to-end MLP to predict a similarity score under both classification and regression objectives. Extensive experiments on five public datasets show that Graph Fusion Model (GFM) achieves state-of-the-art performance and improved efficiency over prior cross-graph interaction methods, with ablations confirming the contribution of each module. The proposed framework provides a scalable, practical approach for graph retrieval and related tasks in real-world graph databases.

Abstract

Graph similarity learning, crucial for tasks such as graph classification and similarity search, focuses on measuring the similarity between two graph-structured entities. The core challenge in this field is effectively managing the interactions between graphs. Traditional methods often entail separate, redundant computations for each graph pair, leading to unnecessary complexity. This paper revolutionizes the approach by introducing a parallel graph interaction method called graph fusion. By merging the node sequences of graph pairs into a single large graph, our method leverages a global attention mechanism to facilitate interaction computations and to harvest cross-graph insights. We further assess the similarity between graph pairs at two distinct levels-graph-level and node-level-introducing two innovative, yet straightforward, similarity computation algorithms. Extensive testing across five public datasets shows that our model not only outperforms leading baseline models in graph-to-graph classification and regression tasks but also sets a new benchmark for performance and efficiency. The code for this paper is open-source and available at https://github.com/LLiRarry/GFM-code.git

Figures (4)

• Figure 1: The overall architecture of our proposed GFM framework
• Figure 2: The overall architecture of our proposed Graph Fusion
• Figure 3: Running time comparisons (average time consumption on one pair of graphs in milliseconds)
• Figure 4: Graph search case study on AIDS dataset