Preprint
Jul 2026
Decoupled Structure-Feature Alignment via Alternating Optimization for Graph Learning
This work presents a constrained two-view learning framework for robust graph learning, which aligns structure-aware GNN embeddings with a structure-free feature prior and designs a channel-split adaptive gated layer within DSAL to effectively integrate this prior.
Cheng-Cheng Yan, Fei-Fei Zhao, Dai-Dai Zhu et al.
· 0 citations