[论文解读] Adversarial Attack and Defense on Graph Data: A Survey
一份综合性综述,统一了图数据上的对抗攻击/防御,回顾了100+篇论文,并为图神经网络及相关方法提供在线资源和分类法。
Deep neural networks (DNNs) have been widely applied to various applications, including image classification, text generation, audio recognition, and graph data analysis. However, recent studies have shown that DNNs are vulnerable to adversarial attacks. Though there are several works about adversarial attack and defense strategies on domains such as images and natural language processing, it is still difficult to directly transfer the learned knowledge to graph data due to its representation structure. Given the importance of graph analysis, an increasing number of studies over the past few years have attempted to analyze the robustness of machine learning models on graph data. Nevertheless, existing research considering adversarial behaviors on graph data often focuses on specific types of attacks with certain assumptions. In addition, each work proposes its own mathematical formulation, which makes the comparison among different methods difficult. Therefore, this review is intended to provide an overall landscape of more than 100 papers on adversarial attack and defense strategies for graph data, and establish a unified formulation encompassing most graph adversarial learning models. Moreover, we also compare different graph attacks and defenses along with their contributions and limitations, as well as summarize the evaluation metrics, datasets and future trends. We hope this survey can help fill the gap in the literature and facilitate further development of this promising new field.
研究动机与目标
- 提供一个关于图数据上对抗攻击与防御工作的广泛、统一视角。
- 引入一个统一的问题表述以比较不同的图对抗方法。
- 建立区分 GNN 与 非-GNN 方法的分类法,并总结数据集与评估指标。
- 突出数据集、指标与未来趋势,以引导基准测试开发和研究方向。
提出的方法
- 提出覆盖污染(poisoning)与规避(evasion)设置的图对抗攻击统一表述(Definitions 3.2)。
- 定义扰动空间 Φ(Gi) 以及通过图距离 Q(·,·) 与预算 ε 来衡量不可感知性(等式 3 与相关讨论)。
- 将攻击与防御分类为 GNN 与 Non-GNN 两大类,并总结策略(边/节点扰动、特征变化、重新连线等)。
- 讨论扰动类型(边级、节点级、结构保持、属性保持)及针对静态与动态图的不可感知性原则。
- 给出一个统一学习框架,对比归纳式与传导式设置(式 1),并扩展到监督/无监督情境。
- 提供一个在线仓库,用于跟踪和更新图对抗学习文献。
实验结果
研究问题
- RQ1有哪些能够涵盖现有图对抗攻击模型的统一表述?
- RQ2如何系统性地对图数据上的对抗攻击与防御进行分类(GNN vs Non-GNN、攻击类型、扰动和任务)?
- RQ3常用的数据集、指标与评估实践有哪些,它们如何支持基准测试开发?
- RQ4在图对抗学习中有哪些关键挑战、局限性与未来方向?
- RQ5如何通过在线资源帮助研究人员跟踪和复现图对抗学习工作?
主要发现
- 该综述覆盖了超过 100 篇论文,提供了图对抗学习的综合格局。
- 提出了用于图对抗攻击的统一表述以统一多种方法(Definitions 3.2)。
- 创建了一个开源在线资源(GitHub),用于跟踪相关工作与代码在图对抗学习中的应用。
- 本文提供了区分基于 GNN 与非 GNN 的攻击/防御的分类法,并总结了评估指标与数据集。
- 它讨论了扰动概念(边/节点扰动、结构/属性保持的变更)以及在静态与动态图中适用的不可感知性原则。
- 工作强调了对比攻击/防御方法的基准测试需求,并概述了未来的研究方向。
更好的研究,从现在开始
从论文设计到论文写作,大幅缩短您的研究时间。
无需绑定信用卡
本解读由 AI 生成,并经人工编辑审核。