Coarse-to-Fine Contrastive Learning on Graphs

• Published in: IEEE transaction on neural networks and learning systems 2024-04, Vol.35 (4), p.4622-4634
• Main Authors: Zhao, Peiyao, Pan, Yuangang, Li, Xin, Chen, Xu, Tsang, Ivor W., Liao, Lejian
• Format: Article
• Language: English
• Subjects: Algorithms; Contrastive learning (CL); Encoding; Feature extraction; graph representation learning; Graphs; Learning; learning to rank (L2R); Mutual information; node representation; Nodes; Perturbation; Perturbation methods; Perturbations; Ranking; Representation learning; self-supervised learning (SSL); Sparse matrices; Task analysis
• ISSN: 2162-237X; 2162-2388
• DOI: 10.1109/TNNLS.2022.3228556

Inspired by the impressive success of contrastive learning (CL), a variety of graph augmentation strategies have been employed to learn node representations in a self-supervised manner. Existing methods construct the contrastive samples by adding perturbations to the graph structure or node attributes. Although impressive results are achieved, it is rather blind to the wealth of prior information assumed: with the increase of the perturbation degree applied on the original graph: 1) the similarity between the original graph and the generated augmented graph gradually decreases and 2) the discrimination between all nodes within each augmented view gradually increases. In this article, we argue that both such prior information can be incorporated (differently) into the CL paradigm following our general ranking framework. In particular, we first interpret CL as a special case of learning to rank (L2R), which inspires us to leverage the ranking order among positive augmented views. Meanwhile, we introduce a self-ranking paradigm to ensure that the discriminative information among different nodes can be maintained and also be less altered to the perturbations of different degrees. Experiment results on various benchmark datasets verify the effectiveness of our algorithm compared with the supervised and unsupervised models.