HomoGCL: Rethinking Homophily in Graph Contrastive Learning
This work addresses a key bottleneck in graph self-supervised learning for researchers and practitioners, offering a plug-and-play improvement that is incremental but effective.
The paper tackles the problem of improving graph contrastive learning by explicitly leveraging homophily, proposing HomoGCL to expand positive sets with neighbor nodes, which achieves state-of-the-art results across six datasets and consistently boosts performance in various methods.
Contrastive learning (CL) has become the de-facto learning paradigm in self-supervised learning on graphs, which generally follows the "augmenting-contrasting" learning scheme. However, we observe that unlike CL in computer vision domain, CL in graph domain performs decently even without augmentation. We conduct a systematic analysis of this phenomenon and argue that homophily, i.e., the principle that "like attracts like", plays a key role in the success of graph CL. Inspired to leverage this property explicitly, we propose HomoGCL, a model-agnostic framework to expand the positive set using neighbor nodes with neighbor-specific significances. Theoretically, HomoGCL introduces a stricter lower bound of the mutual information between raw node features and node embeddings in augmented views. Furthermore, HomoGCL can be combined with existing graph CL models in a plug-and-play way with light extra computational overhead. Extensive experiments demonstrate that HomoGCL yields multiple state-of-the-art results across six public datasets and consistently brings notable performance improvements when applied to various graph CL methods. Code is avilable at https://github.com/wenzhilics/HomoGCL.