Graph Matching Networks for Learning the Similarity of Graph Structured Objects
This addresses the challenge of similarity reasoning for graph-structured data, which is incremental but important for applications like software vulnerability detection.
The paper tackles the problem of retrieving and matching graph-structured objects by training Graph Neural Networks to produce graph embeddings and proposing a Graph Matching Network with cross-graph attention for similarity scoring, demonstrating that it outperforms hand-engineered domain-specific baselines in tasks like control-flow-graph based function similarity search.
This paper addresses the challenging problem of retrieval and matching of graph structured objects, and makes two key contributions. First, we demonstrate how Graph Neural Networks (GNN), which have emerged as an effective model for various supervised prediction problems defined on structured data, can be trained to produce embedding of graphs in vector spaces that enables efficient similarity reasoning. Second, we propose a novel Graph Matching Network model that, given a pair of graphs as input, computes a similarity score between them by jointly reasoning on the pair through a new cross-graph attention-based matching mechanism. We demonstrate the effectiveness of our models on different domains including the challenging problem of control-flow-graph based function similarity search that plays an important role in the detection of vulnerabilities in software systems. The experimental analysis demonstrates that our models are not only able to exploit structure in the context of similarity learning but they can also outperform domain-specific baseline systems that have been carefully hand-engineered for these problems.