Attack-free Evaluating and Enhancing Adversarial Robustness on Categorical Data

Yujun Zhou, Yufei Han, Haomin Zhuang, Hongyan Bao, Xiangliang Zhang
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:61934-61963, 2024.

Abstract

Research on adversarial robustness has predominantly focused on continuous inputs, leaving categorical inputs, especially tabular attributes, less examined. To echo this challenge, our work aims to evaluate and enhance the robustness of classification over categorical attributes against adversarial perturbations through efficient attack-free approaches. We propose a robustness evaluation metric named Integrated Gradient-Smoothed Gradient (IGSG). It is designed to evaluate the attributional sensitivity of each feature and the decision boundary of the classifier, two aspects that significantly influence adversarial risk, according to our theoretical analysis. Leveraging this metric, we develop an IGSG-based regularization to reduce adversarial risk by suppressing the sensitivity of categorical attributes. We conduct extensive empirical studies over categorical datasets of various application domains. The results affirm the efficacy of both IGSG and IGSG-based regularization. Notably, IGSG-based regularization surpasses the state-of-the-art robust training methods by a margin of approximately 0.4% to 12.2% on average in terms of adversarial accuracy, especially on high-dimension datasets. The code is available at https://github.com/YujunZhou/IGSG.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-zhou24i, title = {Attack-free Evaluating and Enhancing Adversarial Robustness on Categorical Data}, author = {Zhou, Yujun and Han, Yufei and Zhuang, Haomin and Bao, Hongyan and Zhang, Xiangliang}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {61934--61963}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/zhou24i/zhou24i.pdf}, url = {https://proceedings.mlr.press/v235/zhou24i.html}, abstract = {Research on adversarial robustness has predominantly focused on continuous inputs, leaving categorical inputs, especially tabular attributes, less examined. To echo this challenge, our work aims to evaluate and enhance the robustness of classification over categorical attributes against adversarial perturbations through efficient attack-free approaches. We propose a robustness evaluation metric named Integrated Gradient-Smoothed Gradient (IGSG). It is designed to evaluate the attributional sensitivity of each feature and the decision boundary of the classifier, two aspects that significantly influence adversarial risk, according to our theoretical analysis. Leveraging this metric, we develop an IGSG-based regularization to reduce adversarial risk by suppressing the sensitivity of categorical attributes. We conduct extensive empirical studies over categorical datasets of various application domains. The results affirm the efficacy of both IGSG and IGSG-based regularization. Notably, IGSG-based regularization surpasses the state-of-the-art robust training methods by a margin of approximately 0.4% to 12.2% on average in terms of adversarial accuracy, especially on high-dimension datasets. The code is available at https://github.com/YujunZhou/IGSG.} }
Endnote
%0 Conference Paper %T Attack-free Evaluating and Enhancing Adversarial Robustness on Categorical Data %A Yujun Zhou %A Yufei Han %A Haomin Zhuang %A Hongyan Bao %A Xiangliang Zhang %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-zhou24i %I PMLR %P 61934--61963 %U https://proceedings.mlr.press/v235/zhou24i.html %V 235 %X Research on adversarial robustness has predominantly focused on continuous inputs, leaving categorical inputs, especially tabular attributes, less examined. To echo this challenge, our work aims to evaluate and enhance the robustness of classification over categorical attributes against adversarial perturbations through efficient attack-free approaches. We propose a robustness evaluation metric named Integrated Gradient-Smoothed Gradient (IGSG). It is designed to evaluate the attributional sensitivity of each feature and the decision boundary of the classifier, two aspects that significantly influence adversarial risk, according to our theoretical analysis. Leveraging this metric, we develop an IGSG-based regularization to reduce adversarial risk by suppressing the sensitivity of categorical attributes. We conduct extensive empirical studies over categorical datasets of various application domains. The results affirm the efficacy of both IGSG and IGSG-based regularization. Notably, IGSG-based regularization surpasses the state-of-the-art robust training methods by a margin of approximately 0.4% to 12.2% on average in terms of adversarial accuracy, especially on high-dimension datasets. The code is available at https://github.com/YujunZhou/IGSG.
APA
Zhou, Y., Han, Y., Zhuang, H., Bao, H. & Zhang, X.. (2024). Attack-free Evaluating and Enhancing Adversarial Robustness on Categorical Data. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:61934-61963 Available from https://proceedings.mlr.press/v235/zhou24i.html.

Related Material