ScaleGraph: A scalable self-supervised framework for cross-domain zero-shot graph learning

Youjiang Fang

doi:10.1016/j.patcog.2025.112482

ScaleGraph: A scalable self-supervised framework for cross-domain zero-shot graph learning

2025-10-03

0

PRE

AI

DOI：10.1016/j.patcog.2025.112482

Original

Original request for help

Save

Abstract

En 中文

• Introduces ScaleGraph, a scalable self-supervised framework for graph representation. • Proposes PGTU, a parameter-free tokenizer aligning heterogeneous node features via kernels. • Develops UniGformer, a unified graphformer with linear attention for large-scale graphs. • Enables dynamic domain adaptation via AdaGraph, a lightweight adaptive graph classifier.

ScaleGraph: A scalable self-supervised framework for cross-domain zero-shot graph learning

Abstract

Journal

Organization