He Mengyao, Zhao Qingqing, Zhang Han
College of Artificial Intelligence, Nankai University, Tongyan Road, Tianjin, 300350, China.
College of Artificial Intelligence, Nankai University, Tongyan Road, Tianjin, 300350, China.
Methods. 2023 Mar;211:31-41. doi: 10.1016/j.ymeth.2023.02.002. Epub 2023 Feb 13.
Self-supervised learning has shown superior performance on graph-related tasks in recent years. The most advanced methods are based on contrast learning, which severely limited by structured data augmentation techniques and complex training methods. Generative self-supervised learning, especially graph autoencoders (GAEs), can prevent the above dependence and has been demonstrated as an effective approach. In addition, most previous works only reconstruct the graph topological structure or node features. Few works consider both and combine them together to obtain their complementary information. To overcome these problems, we propose a generative self-supervised graph representation learning methodology named Multi-View Dual-decoder Graph Autoencoder (MDGA). Specifically, we first design a multi-sample graph learning strategy which benefits the generalization of the dual-decoder graph autoencoder. Moreover, the proposed model reconstructs the graph topological structure with a traditional GAE and extracts node attributes by masked feature reconstruction. Experimental results on five public benchmark datasets demonstrate that MDGA outperforms state-of-the-art methods in both node classification and link prediction tasks.
近年来，自监督学习在与图相关的任务上表现出卓越的性能。最先进的方法基于对比学习，这受到结构化数据增强技术和复杂训练方法的严重限制。生成式自监督学习，尤其是图自动编码器（GAE），可以避免上述依赖，并且已被证明是一种有效的方法。此外，大多数先前的工作仅重建图的拓扑结构或节点特征。很少有工作同时考虑两者并将它们结合起来以获得它们的互补信息。为了克服这些问题，我们提出了一种生成式自监督图表示学习方法，称为多视图双解码器图自动编码器（MDGA）。具体来说，我们首先设计了一种多样本图学习策略，这有利于双解码器图自动编码器的泛化。此外，所提出的模型使用传统的GAE重建图的拓扑结构，并通过掩码特征重建来提取节点属性。在五个公共基准数据集上的实验结果表明，MDGA在节点分类和链接预测任务中均优于当前的先进方法。
