Supervised Graph Contrastive Learning for Few-Shot Node Classification-猫眼课题宝

DOI：

10.1007/978-3-031-26390-3_24

发表时间：

2022-03

期刊：

影响因子：

通讯作者：

中科院分区：

其他

文献类型：

作者：

Zhen Tan;Kaize Ding;Ruocheng Guo;Huan Liu

图存在于许多现实世界的应用中，例如金融欺诈检测、商业推荐和社交网络分析。但是考虑到图注释或标记的高成本，我们面临着严重的图标记稀缺问题，即，一个图可能有几个标记的节点。这样的问题的一个例子是所谓的少数拍摄节点分类。解决这个问题的主要方法是采用情景元学习。在这项工作中，我们通过提出一个基本问题来挑战现状，即元学习是否必须用于少数节点分类任务。我们提出了一个新的和简单的框架下，标准的少数镜头节点分类设置作为替代元学习一个有效的图形编码器。该框架由监督图对比学习与新的机制，数据增强，子图编码和多尺度对比图。在三个基准数据集（CoraFull，Reddit，Ogbn）上的大量实验表明，新框架的性能明显优于最先进的基于元学习的方法。

Graphs present in many real-world applications, such as financial fraud detection, commercial recommendation, and social network analysis. But given the high cost of graph annotation or labeling, we face a severe graph label-scarcity problem, i.e., a graph might have a few labeled nodes. One example of such a problem is the so-calledfew-shot node classification. A predominant approach to this problem resorts toepisodic meta-learning. In this work, we challenge the status quo by asking a fundamental question whether meta-learning is a must for few-shot node classification tasks. We propose a new and simple framework under the standard few-shot node classification setting as an alternative to meta-learning to learn an effective graph encoder. The framework consists of supervised graph contrastive learning with novel mechanisms for data augmentation, subgraph encoding, and multi-scale contrast on graphs. Extensive experiments on three benchmark datasets (CoraFull, Reddit, Ogbn) show that the new framework significantly outperforms state-of-the-art meta-learning based methods.

Supervised Graph Contrastive Learning for Few-Shot Node Classification

文献摘要

相似文献