GraphSAGE++: Weighted Multi-scale GNN for Graph Representation Learning

• DOI: 10.1007/s11063-024-11496-1
• Publication Date: 2024-02-09T10:02:12Z
• Authors: E. Jiawei; Yinglong Zhang; Shangying Yang; Hong Wang; Xuewen Xia; Xing Xu
• Keywords: Representation

Abstract Graph neural networks (GNNs) have emerged as a powerful tool in graph representation learning. However, they are increasingly challenged by over-smoothing network depth grows, compromising their ability to capture and represent complex structures. Additionally, some popular GNN variants only consider local neighbor information during node updating, ignoring the global structural leading inadequate learning differentiation of To address these challenges, we introduce novel framework, GraphSAGE++. Our model extracts target at each layer then concatenates all weighted representations obtain final result. In addition, strategies combining double aggregations with concatenation proposed, which significantly enhance model’s discernment preservation information. Empirical results on various datasets demonstrate that GraphSAGE++ excels vertex classification, link prediction, visualization tasks, surpassing existing methods effectiveness.