arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.07579cs.LG
收藏

TabularBench:面向真实用例中表格深度学习的对抗鲁棒性基准测试

TabularBench: Benchmarking Adversarial Robustness for Tabular Deep Learning in Real-world Use-cases

University of Luxembourg(卢森堡大学);LIST / RIKEN AIP

Thibault Simonetto, Salah Ghamizi, Maxime Cordy

AI总结:

针对表格深度学习对抗鲁棒性研究缺乏标准化基准的问题,提出首个综合基准TabularBench,采用CAA攻击评估,实现7种鲁棒化机制,覆盖多领域200余模型,开源相关库与数据集并给出防御设计见解。

摘要:

尽管计算机视觉领域的对抗鲁棒性已是成熟研究方向,但针对表格深度学习的规避攻击研究较少,对鲁棒化机制与可靠防御的探究则更少。我们假设表格对抗攻击研究的滞后部分源于缺乏标准化基准。为填补这一空白,我们提出TabularBench——首个面向表格深度学习分类模型鲁棒性的综合基准。我们采用CAA(一种梯度与搜索攻击的集成方法,近期被证明是针对表格模型最有效的攻击方法)评估对抗鲁棒性。除了我们开放的基准平台(https://github.com/serval-uni-lu/tabularbench,欢迎提交新模型与防御方法)外,我们还受计算机视觉领域前沿防御启发实现了7种鲁棒化机制,并在金融、医疗与安全领域的5个关键场景中,构建了涵盖200余个模型的最大规模鲁棒表格深度学习基准。我们为每个用例整理了真实数据集,补充了数十万条真实感合成输入,并在有无数据增强的条件下分别训练与评估模型。我们开源了相关库,可通过API访问所有预训练的鲁棒表格模型,以及规模最大的真实与合成表格输入数据集。最后,我们分析了多种防御方法对鲁棒性的影响,为设计新的防御与鲁棒化机制提供了可落地的见解。

英文摘要:

While adversarial robustness in computer vision is a mature research field, fewer researchers have tackled the evasion attacks against tabular deep learning, and even fewer investigated robustification mechanisms and reliable defenses. We hypothesize that this lag in the research on tabular adversarial attacks is in part due to the lack of standardized benchmarks. To fill this gap, we propose TabularBench, the first comprehensive benchmark of robustness of tabular deep learning classification models. We evaluated adversarial robustness with CAA, an ensemble of gradient and search attacks which was recently demonstrated as the most effective attack against a tabular model. In addition to our open benchmark (https://github.com/serval-uni-lu/tabularbench) where we welcome submissions of new models and defenses, we implement 7 robustification mechanisms inspired by state-of-the-art defenses in computer vision and propose the largest benchmark of robust tabular deep learning over 200 models across five critical scenarios in finance, healthcare and security. We curated real datasets for each use case, augmented with hundreds of thousands of realistic synthetic inputs, and trained and assessed our models with and without data augmentations. We open-source our library that provides API access to all our pre-trained robust tabular models, and the largest datasets of real and synthetic tabular inputs. Finally, we analyze the impact of various defenses on the robustness and provide actionable insights to design new defenses and robustification mechanisms.

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。

↑