TabularBench
收藏资源简介:
TabularBench是由卢森堡大学创建的一个综合基准数据集,专门用于评估表格深度学习分类模型的对抗鲁棒性。该数据集包含五个关键领域(金融、医疗保健和安全等)的真实和合成数据,数据量达到数十万条。数据集的创建过程结合了最新的数据增强技术和对抗训练方法,旨在解决表格数据在对抗攻击下的鲁棒性问题。TabularBench的应用领域广泛,特别是在需要高度安全性和准确性的机器学习模型中,如金融评分和医疗诊断。
TabularBench is a comprehensive benchmark dataset developed by the University of Luxembourg, specifically designed to evaluate the adversarial robustness of tabular deep learning classification models. This dataset includes real and synthetic data from five key domains such as finance, healthcare and security, with a total of hundreds of thousands of samples. The dataset was constructed using state-of-the-art data augmentation techniques and adversarial training methods, aiming to address the adversarial robustness issues of tabular data under adversarial attacks. TabularBench has broad application scenarios, particularly in machine learning models requiring high security and accuracy, such as financial scoring and medical diagnosis.




