Jul 2026
Unsupervised Graph Representation Learning with Complementary View Alignment
This framework introduces a dual-encoder architecture that separately processes structural and attribute information, incorporates node positional encoding to approximate Neighborhood Identity Distribution (NID), and employs dual reconstruction tasks for both edges and node attributes.
Zengyi Wo, Shiyu Zhang, Qiyao Peng et al.
· arXiv.org · 0 citations