From One to All: Learning to Match Heterogeneous and Partially Overlapped Graphs

TL;DR

This paper introduces a novel unsupervised learning method for matching heterogeneous and partially overlapping graphs, leveraging a partial optimal transport paradigm to improve alignment accuracy in complex real-world scenarios.

Contribution

It presents the first practical learning-to-match approach that handles heterogeneous, partially overlapped graphs using a from-one-to-all strategy and a novel partial OT framework.

Findings

Outperforms existing graph matching methods in experiments.

Effectively handles heterogeneous and partially overlapping graphs.

Introduces a new partial OT-based learning paradigm.

Abstract

Recent years have witnessed a flurry of research activity in graph matching, which aims at finding the correspondence of nodes across two graphs and lies at the heart of many artificial intelligence applications. However, matching heterogeneous graphs with partial overlap remains a challenging problem in real-world applications. This paper proposes the first practical learning-to-match method to meet this challenge. The proposed unsupervised method adopts a novel partial OT paradigm to learn a transport plan and node embeddings simultaneously. In a from-one-to-all manner, the entire learning procedure is decomposed into a series of easy-to-solve sub-procedures, each of which only handles the alignment of a single type of nodes. A mechanism for searching the transport mass is also proposed. Experimental results demonstrate that the proposed method outperforms state-of-the-art graph matching methods.