Monotone Adversary

Monotone adversaries represent a challenging class of attackers in machine learning, focusing on scenarios where malicious actors strategically manipulate data or models to achieve a specific goal, such as misclassifying inputs or degrading model performance, while maintaining a consistent pattern of malicious behavior. Current research investigates robust defenses against these adversaries across various machine learning tasks, including federated learning, reinforcement learning, and natural language processing, employing techniques like list-decodable learning, activation space analysis, and adaptive budget allocation in multi-agent systems. Understanding and mitigating the impact of monotone adversaries is crucial for building reliable and secure machine learning systems across diverse applications, from autonomous systems to large language models.

Papers

November 1, 2024

Learning in Markov Games with Adaptive Adversaries: Policy Regret, Fundamental Barriers, and Efficient Algorithms
Thanh Nguyen-Tang, Raman Arora
LeArning Abstract Markov Game Sample Efficient Efficient Algorithm Adaptive Adversary Monotone Adversary Counterfactual Regret

August 9, 2024

LiD-FL: Towards List-Decodable Federated Learning
Hong Liu, Liren Shan, Han Bao, Ronghui You, Yuhao Yi, Jiancheng Lv
Adversarial Attack Convex Loss Monotone Adversary

July 21, 2024

Mitigating Deep Reinforcement Learning Backdoors in the Neural Activation Space
Sanyam Vyas, Chris Hicks, Vasilios Mavroudis
Deep Reinforcement Learning Backdoor Attack Backdoor Trigger Activation Space Monotone Adversary

June 28, 2024

Covert Malicious Finetuning: Challenges in Safeguarding LLM Adaptation
Danny Halawi, Alexander Wei, Eric Wallace, Tony T. Wang, Nika Haghtalab, Jacob Steinhardt
Fine Tuning Technical Challenge LLM Adaptation Monotone Adversary Malicious Sample Black Box Tuning

May 1, 2024

Adversarial Attacks and Defense for Conversation Entailment Task
Zhenning Yang, Ryan Krawec, Liang-Yuan Wu
Large Language Model Adversarial Attack Textual Entailment Fast Adversarial Training Monotone Adversary

February 20, 2024

EvoGrad: A Dynamic Take on the Winograd Schema Challenge with Human Adversaries
Jing Han Sun, Ali Emami
New Perspective Commonsense Reasoning Task Monotone Adversary Winograd Convolution Pronoun Translation

February 12, 2024

Top-$K$ ranking with a monotone adversary
Yuepeng Yang, Antares Chen, Lorenzo Orecchia, Cong Ma
Sample Complexity Pairwise Comparison Random Graph Monotone Adversary

November 18, 2023

Dueling Optimization with a Monotone Adversary
Avrim Blum, Meghal Gupta, Gene Li, Naren Sarayu Manoj, Aadirupa Saha, Yuanyuan Yang
Optimization Purpose Convex Optimization Online Algorithm Randomized Algorithm Monotone Adversary

June 27, 2023

Cooperation or Competition: Avoiding Player Domination for Multi-Target Robustness via Adaptive Budgets
Yimu Wang, Dinghuai Zhang, Yihan Wu, Heng Huang, Hongyang Zhang
Adversarial Attack Long Term Cooperation Competition Platform Adversary Agent Monotone Adversary Robust Network Double Robustness Matrix Domination

April 25, 2023

Combining Adversaries with Anti-adversaries in Training
Xiaoling Zhou, Nan Yang, Ou Wu
Training Data Adversarial Training Adversarial Perturbation Multi Step Adversarial Attack Monotone Adversary Standard Adversarial Training Imbalance Learning Adversarial Direction

February 13, 2023

Achieving Better Regret against Strategic Adversaries
Le Cong Dinh, Tri-Dung Nguyen, Alain Zemkoho, Long Tran-Thanh
O$ Regret Online Learning Algorithm Low Regret Monotone Adversary Counterfactual Regret

February 11, 2023

Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attacks
Daniel Kang, Xuechen Li, Ion Stoica, Carlos Guestrin, Matei Zaharia, Tatsunori Hashimoto
Large Language Model Instruction Following Instruction Tuned Large Language Model Monotone Adversary Malicious Content Software Behavior

December 20, 2022

AdverSAR: Adversarial Search and Rescue via Multi-Agent Reinforcement Learning
Aowabin Rahman, Arnab Bhattacharya, Thiagarajan Ramachandran, Sayak Mukherjee, Himanshu Sharma, Ted Fujimoto, Samrat Chatterjee
Multi Agent Reinforcement Learning Search and Rescue Multi Robot Team Monotone Adversary Adversarial Search Adversarial Multi Adversarial Communication

October 13, 2022

Observed Adversaries in Deep Reinforcement Learning
Eugene Lim, Harold Soh
Adversarial Attack Deep Reinforcement Learning Adversarial Perspective Monotone Adversary Deep Policy Adversary Knowledge

August 21, 2022

Inferring Sensitive Attributes from Model Explanations
Vasisht Duddu, Antoine Boutet
Model Explanation Sensitive Attribute Attribute Inference Attack Monotone Adversary

August 10, 2022

Reducing Exploitability with Population Based Training
Pavel Czempin, Adam Gleave
Adversarial Training Zero Sum Game Adversarial Policy Monotone Adversary Population Based Training Self Play Reinforcement Learning

April 28, 2022

Adversarial Fine-tune with Dynamically Regulated Adversary
Pengyue Hou, Ming Zhou, Jie Han, Petr Musilek, Xingyu Li
Adversarial Attack Adversarial Training Adversarial Robustness Adversarial Fine Tuning Monotone Adversary Adversarial Training Algorithm

February 28, 2022

Distributed randomized Kaczmarz for the adversarial workers
Xia Li, Longxiu Huang, Deanna Needell
Least Square Adversary Agent Adversarial Agent Iterative Approach Monotone Adversary Kaczmarz Algorithm

January 13, 2022

Reconstructing Training Data with Informed Adversaries
Borja Balle, Giovanni Cherubin, Jamie Hayes
Training Data Reconstruction Attack High Fidelity Reconstruction Monotone Adversary Adversary Knowledge