A New Ensemble Adversarial Attack Powered by Long-term Gradient Memories

TL;DR

This paper introduces a novel ensemble adversarial attack method that leverages long-term gradient memories to improve attack effectiveness against deep neural networks.

Contribution

It presents a new ensemble attack technique utilizing long-term gradient memories, advancing the state-of-the-art in adversarial attack strategies.

Findings

Enhanced attack success rates on multiple neural network models

Demonstrated robustness of the attack against defenses

Improved transferability of adversarial examples

Abstract

Deep neural networks are vulnerable to adversarial attacks.