Robust graph contrastive learning with multi-hop views for node classification

Yutong Wang, Junheng Zhang, Ren Cao, Minhao Zou, Chun Guan, Siyang Leng

Abstract: Highlights•We propose MHVGCL, a graph contrastive framework with two versions for reduced overhead.•We design an augmentation method for graph data, generating views with multi-hop information.•We propose a multi-hop contrastive loss to leverage multi-hop views.