Jul 2026
What Makes Graph Unified? Principles and Generative Sliding-Window Transformer for Graph Foundation Models
SliGFM is proposed, a graph foundation model built upon topology-aware sliding-window feature encoding and generative reconstruction that enables a smoothness-aware transformer to capture transferable relational patterns among feature tokens within each node, while the generative reconstruction objective encourages preservation of the original feature information.
Dongxiao He, Siqi Liu, Jitao Zhao et al.
· arXiv.org · 0 citations