首个针对表格数据的对抗攻击基准，评估攻击效果与隐蔽性平衡。
TabAttackBench: A Benchmark for Adversarial Attacks on Tabular Data
- 构建多维度评估框架，量化攻击在表格数据中的隐蔽性与有效性。
- $\ℓ_\infty$攻击成功率高但痕迹明显，$\ℓ_2$攻击更逼真。
- 适用于关注表格模型安全性的研究人员和工业应用开发者。
对抗攻击通过微小扰动诱导机器学习模型产生错误预测，威胁显著。尽管图像等非结构化数据上的攻击研究成熟，表格数据因特征类型混合及复杂依赖关系，相关研究仍不足。本文提出一个综合性基准，评估表格数据上对抗攻击的有效性与隐蔽性。在金融、能源、医疗领域的十一组数据集上，测试了五种白盒攻击算法（FGSM、BIM、PGD、DeepFool、C&W）在四种模型（LR、MLP、TabTransformer、FT-Transformer）上的表现。采用四类定量隐蔽性指标（接近度、稀疏性、偏离度、敏感性）刻画扰动真实性。结果揭示：$\ℓ_\infty$攻击成功率更高但更易察觉，$\ℓ_2$攻击更具现实感。该基准为设计更隐蔽的攻击提供实践指导，深化对表格机器学习模型脆弱性的理解。
原文摘要 · Abstract (English)
Adversarial attacks pose a significant threat to machine learning models by inducing incorrect predictions through imperceptible perturbations to input data. While these attacks are well studied in unstructured domains such as images, their behaviour on tabular data remains underexplored due to mixed feature types and complex inter-feature dependencies. This study introduces a comprehensive benchmark that evaluates adversarial attacks on tabular datasets with respect to both effectiveness and imperceptibility. We assess five white-box attack algorithms (FGSM, BIM, PGD, DeepFool, and C\&W) across four representative models (LR, MLP, TabTransformer and FT-Transformer) using eleven datasets spanning finance, energy, and healthcare domains. The benchmark employs four quantitative imperceptibility metrics (proximity, sparsity, deviation, and sensitivity) to characterise perturbation realism. The analysis quantifies the trade-off between these two aspects and reveals consistent differences between attack types, with $\ell_\infty$-based attacks achieving higher success but lower subtlety, and $\ell_2$-based attacks offering more realistic perturbations. The benchmark findings offer actionable insights for designing more imperceptible adversarial attacks, advancing the understanding of adversarial vulnerability in tabular machine learning.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。