GCN-SE: Attention as Explainability for Node Classification in Dynamic Graphs
GCN-SE: Attention as Explainability for Node Classification in Dynamic Graphs
复制标题
DOI：
10.1109/icdm51629.2021.00123
复制
发表时间：
2021-10
期刊：
影响因子：
--
通讯作者：
Yucai Fan;Yuhang Yao;Carlee Joe-Wong
中科院分区：
文献类型：
--
作者：
Yucai Fan;Yuhang Yao;Carlee Joe-Wong
图卷积网络（GCN）是一种流行的图表示学习方法，已被证明对节点分类等任务有效。传统GCN模型的最新变体旨在对拓扑和节点属性随时间变化的动态图中的节点进行分类，例如，具有动态关系的社交网络。然而，这些工作并没有完全解决在不同时间灵活地向图的快照分配不同重要性的挑战，这取决于图动态可能对标签具有或多或少的预测能力。我们通过提出一种新的方法GCN-SE来解决这一挑战，该方法在不同时间将一组可学习的注意力权重附加到图快照上，其灵感来自挤压和激励网络（SE-Net）。我们表明，GCNSE优于以前提出的节点分类方法在各种图形数据集。为了验证注意力权重在确定不同图快照重要性方面的有效性，我们将可解释机器学习领域的基于扰动的方法应用于图形设置，并评估GCN-SE学习的注意力权重与不同快照重要性之间的相关性。
Graph Convolutional Networks (GCNs) are a popular method from graph representation learning that have proved effective for tasks like node classification. Recent variants on traditional GCN models aim to classify nodes in dynamic graphs whose topologies and node attributes change over time, e.g., social networks with dynamic relationships. These works, however, do not fully address the challenge of flexibly assigning different importance to snapshots of the graph at different times, which depending on the graph dynamics may have more or less predictive power on the labels. We address this challenge by proposing a new method, GCN-SE, that attaches a set of learnable attention weights to graph snapshots at different times, inspired by Squeeze and Excitation Net (SE-Net). We show that GCNSE outperforms previously proposed node classification methods on a variety of graph datasets. To verify the effectiveness of the attention weight in determining the importance of different graph snapshots, we adapt perturbation-based methods from the field of explainable machine learning to graphical settings and evaluate the correlation between the attention weights learned by GCN-SE and the importance of different snapshots over time.