Multiple Adversary

Multiple adversary research focuses on enhancing the robustness of machine learning models, particularly deep neural networks, against a variety of simultaneous attacks. Current research explores diverse attack strategies, including multi-granular perturbations, Bayesian approaches, and multi-trigger backdoors, and employs models like generative adversarial networks and reinforcement learning to generate and defend against these attacks. This work is crucial for improving the security and reliability of AI systems in safety-critical applications, addressing vulnerabilities that could lead to catastrophic failures or malicious exploitation. The development of robust defense mechanisms is a key objective, with recent efforts focusing on interpreter-based ensembles and uncertainty-aware techniques.

Papers

October 21, 2024

A Troublemaker with Contagious Jailbreak Makes Chaos in Honest Towns
Tianyi Men, Pengfei Cao, Zhuoran Jin, Yubo Chen, Kang Liu, Jun Zhao
Multi Agent Multi Agent System Jailbreak Attack Kill Chaos Truthful Space Sex Trouble Multiple Adversary

August 20, 2024

Revisiting Min-Max Optimization Problem in Adversarial Training
Sina Hajer Ahmadi, Hassan Bahrami
Adversarial Attack Adversarial Example Adversarial Training Robust Deep Min Max Multiple Adversary

August 14, 2024

Robust Active Learning (RoAL): Countering Dynamic Adversaries in Active Learning with Elastic Weight Consolidation
Ricky Maulana Fajri, Yulong Pei, Lu Yin, Mykola Pechenizkiy
Adversarial Attack Active Learning Multi Step Adversarial Attack Weight Consolidation Multiple Adversary

July 11, 2024

How to beat a Bayesian adversary
Zihan Ding, Kexin Jin, Jonas Latz, Chenguang Liu
Adversarial Attack Gradient Flow Robust Machine Learning Multiple Adversary

April 20, 2024

Generative Subspace Adversarial Active Learning for Outlier Detection in Multiple Views of High-dimensional Data
Jose Cribeiro-Ramallo, Vadim Arzamasov, Federico Matteucci, Denis Wambold, Klemens Böhm
Generative Adversarial Network Outlier Detection High Dimensional Data Multiple View Unsupervised Outlier Detection Multiple Adversary Generative Active Learning

April 2, 2024

Multi-granular Adversarial Attacks against Black-box Neural Ranking Models
Yu-An Liu, Ruqing Zhang, Jiafeng Guo, Maarten de Rijke, Yixing Fan, Xueqi Cheng
Adversarial Text Neural Ranker Multiple Adversary

January 27, 2024

Shortcuts Everywhere and Nowhere: Exploring Multi-Trigger Backdoor Attacks
Yige Li, Jiabo He, Hanxun Huang, Jun Sun, Xingjun Ma
Backdoor Attack Backdoor Data Trigger Attack Multiple Adversary

April 14, 2023

Interpretability is a Kind of Safety: An Interpreter-based Ensemble for Adversary Defense
Jingyuan Wang, Yufan Wu, Mingxuan Li, Xin Lin, Junjie Wu, Chao Li
Adversarial Attack Adversarial Example Inherent Interpretability Human SAFETY Attack Robustness Multiple Adversary

November 8, 2022

NaturalAdversaries: Can Naturalistic Adversaries Be as Effective as Artificial Adversaries?
Saadia Gabriel, Hamid Palangi, Yejin Choi
Adversarial Example Multi Step Adversarial Attack White Box Adversarial Attack Previously Generated Alternative Neural Text Classification Multiple Adversary

May 4, 2022

Uncertainty-Autoencoder-Based Privacy and Utility Preserving Data Type Conscious Transformation
Bishwas Mandal, George Amariucai, Shuangqing Wei
Privacy Utility Privacy Mechanism Preserving Transformation Multiple Adversary Private Variational

Multiple Adversary

Papers

A Troublemaker with Contagious Jailbreak Makes Chaos in Honest Towns

Revisiting Min-Max Optimization Problem in Adversarial Training

Robust Active Learning (RoAL): Countering Dynamic Adversaries in Active Learning with Elastic Weight Consolidation

How to beat a Bayesian adversary

Generative Subspace Adversarial Active Learning for Outlier Detection in Multiple Views of High-dimensional Data

Multi-granular Adversarial Attacks against Black-box Neural Ranking Models

Shortcuts Everywhere and Nowhere: Exploring Multi-Trigger Backdoor Attacks

Interpretability is a Kind of Safety: An Interpreter-based Ensemble for Adversary Defense

NaturalAdversaries: Can Naturalistic Adversaries Be as Effective as Artificial Adversaries?

Uncertainty-Autoencoder-Based Privacy and Utility Preserving Data Type Conscious Transformation