arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.04710cs.LG
收藏

结合基于检索的因果学习与信息瓶颈的可解释图神经网络

Incorporating Retrieval-based Causal Learning with Information Bottlenecks for Interpretable Graph Neural Networks

School of Computer Science and Engineering, Sun Yat-sen University(中山大学计算机科学与工程学院);Shanghai Jiao Tong University(上海交通大学);Key Laboratory of Machine Intelligence and Advanced Computing, Sun Yat-sen University(中山大学机器智能与先进计算重点实验室)

Jiahua Rao, Jiancong Xie, Hanjing Lin, Shuangjia Zheng, Zhen Wang, Yuedong Yang

AI总结:

本研究提出一种结合检索式因果学习与图信息瓶颈的可解释因果GNN框架,能半参数化检索关键子图并压缩解释,在提升解释精度的同时增强GNN预测性能。

摘要:

图神经网络(GNNs)因其有效处理拓扑数据的能力而受到广泛关注,但其可解释性仍是一个关键问题。当前的解释方法主要采用事后解释,以提供对GNNs透明且直观的理解。然而,这些方法在解释复杂子图方面性能有限,且无法利用解释来提升GNN的预测能力。另一方面,透明GNN模型被提出用于捕获关键子图。尽管此类方法能提升GNN预测,但通常在解释方面表现不佳。因此,需要一种新策略来更好地结合GNN的解释与预测。在本研究中,我们开发了一种新颖的可解释因果GNN框架,该框架将基于检索的因果学习与图信息瓶颈(GIB)理论相结合。该框架能够半参数化地检索由GIB检测到的关键子图,并通过因果模块压缩解释性子图。实验表明,该框架在多种解释类型的真实世界解释场景中持续优于最先进的方法,精度提高了32.71%。更重要的是,学习到的解释被证明也能提升GNN的预测性能。

英文摘要:

Graph Neural Networks (GNNs) have gained considerable traction for their capability to effectively process topological data, yet their interpretability remains a critical concern. Current interpretation methods are dominated by post-hoc explanations to provide a transparent and intuitive understanding of GNNs. However, they have limited performance in interpreting complicated subgraphs and can't utilize the explanation to advance GNN predictions. On the other hand, transparent GNN models are proposed to capture critical subgraphs. While such methods could improve GNN predictions, they usually don't perform well on explanations. Thus, it is desired for a new strategy to better couple GNN explanation and prediction. In this study, we have developed a novel interpretable causal GNN framework that incorporates retrieval-based causal learning with Graph Information Bottleneck (GIB) theory. The framework could semi-parametrically retrieve crucial subgraphs detected by GIB and compress the explanatory subgraphs via a causal module. The framework was demonstrated to consistently outperform state-of-the-art methods, and to achieve 32.71\% higher precision on real-world explanation scenarios with diverse explanation types. More importantly, the learned explanations were shown able to also improve GNN prediction performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑