Pareto Adversarial Robustness: Balancing Spatial Robustness and Sensitivity-based Robustness

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Sun, Ke, Li, Mingjie, Lin, Zhouchen
Formato: Preprint
Publicado: 2021
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910499103309824
author Sun, Ke
Li, Mingjie
Lin, Zhouchen
author_facet Sun, Ke
Li, Mingjie
Lin, Zhouchen
contents Adversarial robustness, which primarily comprises sensitivity-based robustness and spatial robustness, plays an integral part in achieving robust generalization. In this paper, we endeavor to design strategies to achieve universal adversarial robustness. To achieve this, we first investigate the relatively less-explored realm of spatial robustness. Then, we integrate the existing spatial robustness methods by incorporating both local and global spatial vulnerability into a unified spatial attack and adversarial training approach. Furthermore, we present a comprehensive relationship between natural accuracy, sensitivity-based robustness, and spatial robustness, supported by strong evidence from the perspective of robust representation. Crucially, to reconcile the interplay between the mutual impacts of various robustness components into one unified framework, we incorporate the \textit{Pareto criterion} into the adversarial robustness analysis, yielding a novel strategy called Pareto Adversarial Training for achieving universal robustness. The resulting Pareto front, which delineates the set of optimal solutions, provides an optimal balance between natural accuracy and various adversarial robustness. This sheds light on solutions for achieving universal robustness in the future. To the best of our knowledge, we are the first to consider universal adversarial robustness via multi-objective optimization.
format Preprint
id arxiv_https___arxiv_org_abs_2111_01996
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Pareto Adversarial Robustness: Balancing Spatial Robustness and Sensitivity-based Robustness
Sun, Ke
Li, Mingjie
Lin, Zhouchen
Machine Learning
Adversarial robustness, which primarily comprises sensitivity-based robustness and spatial robustness, plays an integral part in achieving robust generalization. In this paper, we endeavor to design strategies to achieve universal adversarial robustness. To achieve this, we first investigate the relatively less-explored realm of spatial robustness. Then, we integrate the existing spatial robustness methods by incorporating both local and global spatial vulnerability into a unified spatial attack and adversarial training approach. Furthermore, we present a comprehensive relationship between natural accuracy, sensitivity-based robustness, and spatial robustness, supported by strong evidence from the perspective of robust representation. Crucially, to reconcile the interplay between the mutual impacts of various robustness components into one unified framework, we incorporate the \textit{Pareto criterion} into the adversarial robustness analysis, yielding a novel strategy called Pareto Adversarial Training for achieving universal robustness. The resulting Pareto front, which delineates the set of optimal solutions, provides an optimal balance between natural accuracy and various adversarial robustness. This sheds light on solutions for achieving universal robustness in the future. To the best of our knowledge, we are the first to consider universal adversarial robustness via multi-objective optimization.
title Pareto Adversarial Robustness: Balancing Spatial Robustness and Sensitivity-based Robustness
topic Machine Learning
url https://arxiv.org/abs/2111.01996