Constructing Semantics-Aware Adversarial Examples with a Probabilistic Perspective

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Andi, Zhang, Mingtian, Wischik, Damon
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916493323665408
author Zhang, Andi
Zhang, Mingtian
Wischik, Damon
author_facet Zhang, Andi
Zhang, Mingtian
Wischik, Damon
contents We propose a probabilistic perspective on adversarial examples, allowing us to embed subjective understanding of semantics as a distribution into the process of generating adversarial examples, in a principled manner. Despite significant pixel-level modifications compared to traditional adversarial attacks, our method preserves the overall semantics of the image, making the changes difficult for humans to detect. This extensive pixel-level modification enhances our method's ability to deceive classifiers designed to defend against adversarial attacks. Our empirical findings indicate that the proposed methods achieve higher success rates in circumventing adversarial defense mechanisms, while remaining difficult for human observers to detect.
format Preprint
id arxiv_https___arxiv_org_abs_2306_00353
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Constructing Semantics-Aware Adversarial Examples with a Probabilistic Perspective
Zhang, Andi
Zhang, Mingtian
Wischik, Damon
Machine Learning
Cryptography and Security
We propose a probabilistic perspective on adversarial examples, allowing us to embed subjective understanding of semantics as a distribution into the process of generating adversarial examples, in a principled manner. Despite significant pixel-level modifications compared to traditional adversarial attacks, our method preserves the overall semantics of the image, making the changes difficult for humans to detect. This extensive pixel-level modification enhances our method's ability to deceive classifiers designed to defend against adversarial attacks. Our empirical findings indicate that the proposed methods achieve higher success rates in circumventing adversarial defense mechanisms, while remaining difficult for human observers to detect.
title Constructing Semantics-Aware Adversarial Examples with a Probabilistic Perspective
topic Machine Learning
Cryptography and Security
url https://arxiv.org/abs/2306.00353