Welcome to TabularBench’s documentation!
TabularBench is a comprehensive benchmark of robustness of tabular deep learning classification models. The benchmark implements 3 Tabular attacks: MOEVa, CAPGD and CAA. And support 5 datasets, 5 tabular model architectures and 7 data augmentation mechanisms.
The benchmark provides pre-processed constrained datasets, as well as pre-trained robust tabular models. The results of the benchmark can be found on [TabularBench website](https://serval-uni-lu.github.io/tabularbench/)
Getting started
API Reference
How to cite?
@article{simonetto2024constrained,
title={Constrained Adaptive Attack: Effective Adversarial Attack Against Deep Neural Networks for Tabular Data},
author={Simonetto, Thibault and Ghamizi, Salah and Cordy, Maxime},
journal={arXiv preprint arXiv:2406.00775},
year={2024}
}