This website requires JavaScript.

Multi-duplicated Characterization of Graph Structures using Information Gain Ratio for Graph Neural Networks

Y. OishiK. kaneiwa
Dec 2022
摘要
Various graph neural networks (GNNs) have been proposed to solve nodeclassification tasks in machine learning for graph data. GNNs use thestructural information of graph data by aggregating the features of neighboringnodes. However, they fail to directly characterize and leverage the structuralinformation. In this paper, we propose multi-duplicated characterization ofgraph structures using information gain ratio (IGR) for GNNs (MSI-GNN), whichenhances the performance of node classification by using an i-hop adjacencymatrix as the structural information of the graph data. In MSI-GNN, the i-hopadjacency matrix is adaptively adjusted by two methods: (i) structural featuresin the matrix are selected based on the IGR, and (ii) the selected features in(i) for each node are duplicated and combined flexibly. In an experiment, weshow that our MSI-GNN outperforms GCN, H2GCN, and GCNII in terms of averageaccuracies in benchmark graph datasets.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答