Higher -order expanded heterogeneous graph framework for attribute completion with bi-level programming

doi:10.1016/j.patcog.2025.112936

Higher -order expanded heterogeneous graph framework for attribute completion with bi-level programming

2025-12-25

0

PRE

AI

蒋怀光 (Huaiguang Jiang)

DOI：10.1016/j.patcog.2025.112936

Original

Original request for help

Save

Abstract

En 中文

• We propose a higher-order expansion mechanism to enrich node information and improve attribute imputation. • Attribute completion and heterogeneous graph parameter learning are formulated as a bi-level optimization problem, enhancing generalization via downstream validation loss. • Experiments on three real-world datasets show our framework outperforms state-of-the-art in heterogeneous graph attribute completion.

Higher -order expanded heterogeneous graph framework for attribute completion with bi-level programming

Abstract

Journal

Organization

Cited Papers