2,164 Hits in 11.1 sec

Sparse Adversarial Attack in Multi-agent Reinforcement Learning [article]

Yizheng Hu, Zhihua Zhang
2022 arXiv   pre-print
Cooperative multi-agent reinforcement learning (cMARL) has many real applications, but the policy trained by existing cMARL algorithms is not robust enough when deployed.  ...  In this paper, we propose a sparse adversarial attack on cMARL systems. We use (MA)RL with regularization to train the attack policy.  ...  Conclusion and Future works In this work we have been concerned with sparse attack in multi agent reinforcement learning.  ... 
arXiv:2205.09362v1 fatcat:n7hawt5ieffzhav6wt2zgw2n5y

Reinforcement Learning Based Sparse Black-box Adversarial Attack on Video Recognition Models [article]

Zeyuan Wang, Chaofeng Sha, Su Yang
2021 arXiv   pre-print
In order to speed up the attack process, we propose a reinforcement learning based frame selection strategy.  ...  We explore the black-box adversarial attack on video recognition models.  ...  Reinforcement Learning (RL) has strong potential in sequence decision making.  ... 
arXiv:2108.13872v1 fatcat:m4fpyjlianhkfa6juynzcs72ta

A Generative Model based Adversarial Security of Deep Learning and Linear Classifier Models [article]

erhat Ozgur Catak and Samed Sivaslioglu and Kevser Sahinbas
2020 arXiv   pre-print
In this paper, we have proposed a mitigation method for adversarial attacks against machine learning models with an autoencoder model that is one of the generative ones.  ...  The main idea behind adversarial attacks against machine learning models is to produce erroneous results by manipulating trained models.  ...  They propose a "Robust Adversarial Reinforced Learning" (RARL), where they train an agent to operate in the presence of a destabilizing adversary that applies disturbance forces to the system.  ... 
arXiv:2010.08546v1 fatcat:trqowc5b5jbnvaqvgafyiui76m

Fixed Points in Cyber Space: Rethinking Optimal Evasion Attacks in the Age of AI-NIDS [article]

Christian Schroeder de Witt, Yongchao Huang, Philip H.S. Torr, Martin Strohmeier
2021 arXiv   pre-print
We then argue that attacker-defender fixed points are themselves general-sum games with complex phase transitions, and introduce a temporally extended multi-agent reinforcement learning framework in which  ...  Instead, we suggest a novel reinforcement learning setting that can be used to efficiently generate arbitrary adversarial perturbations.  ...  Adversarial blackbox attacks on network flow classifiers have recently been demonstrated using reinforcement learning with a sparse feedback signal in order to generate adversarial perturbations allowing  ... 
arXiv:2111.12197v1 fatcat:gqo27ytqizcdrfsbiicfsba37m

Failure-Scenario Maker for Rule-Based Agent using Multi-agent Adversarial Reinforcement Learning and its Application to Autonomous Driving [article]

Akifumi Wachi
2019 arXiv   pre-print
We examine the problem of adversarial reinforcement learning for multi-agent domains including a rule-based agent.  ...  We propose a method for efficiently finding failure scenarios; this method trains the adversarial agents using multi-agent reinforcement learning such that the tested rule-based agent fails.  ...  Multi-agent reinforcement learning (MARL) The relationship among multiple agents can be categorized into cooperative, competitive, and both.  ... 
arXiv:1903.10654v3 fatcat:jumtvfjvgjcpxaom6rxxfrl42q

Multi-agent Reinforcement Learning in Bayesian Stackelberg Markov Games for Adaptive Moving Target Defense [article]

Sailik Sengupta, Subbarao Kambhampati
2020 arXiv   pre-print
or tend to ignore the strategic nature of the adversary simplifying the scenario to use single-agent reinforcement learning techniques.  ...  to an optimal policy in MTD domains with incomplete information about adversaries even when prior information about rewards and transitions is absent.  ...  Acknowlegements The research is supported in part by ONR grants N00014-16-1-2892, N00014-18-1-2442, N00014-18-1-2840, N00014-19-1-2119, AFOSR grant FA9550-18-1-0067, DARPA SAIL-ON grant W911NF-19-2-0006  ... 
arXiv:2007.10457v1 fatcat:ddz3g2cezza57lhf3tdhtdkwxa

Multi-Agent Vulnerability Discovery for Autonomous Driving with Hazard Arbitration Reward [article]

Weilin Liu, Ye Mu, Chao Yu, Xuefei Ning, Zhong Cao, Yi Wu, Shuang Liang, Huazhong Yang, Yu Wang
2021 arXiv   pre-print
To this end, this work proposes a Safety Test framework by finding Av-Responsible Scenarios (STARS) based on multi-agent reinforcement learning.  ...  Discovering hazardous scenarios is crucial in testing and further improving driving policies. However, conducting efficient driving policy testing faces two key challenges.  ...  Specifically, we introduce Multi-Agent Reinforcement Learning (MARL) to control other traffic participants to interact with the under-test driving policy adversarially.  ... 
arXiv:2112.06185v1 fatcat:fdava3cuyzezdksbqeifo3ohwm

Towards Closing the Sim-to-Real Gap in Collaborative Multi-Robot Deep Reinforcement Learning [article]

Wenshuai Zhao, Jorge Peña Queralta, Li Qingqing, Tomi Westerlund
2020 arXiv   pre-print
Current research directions in deep reinforcement learning include bridging the simulation-reality gap, improving sample efficiency of experiences in distributed multi-agent reinforcement learning, together  ...  with the development of robust methods against adversarial agents in distributed learning, among many others.  ...  The literature in adversarial conditions for collaborative multi-agent learning is, nonetheless, sparse. Adversarial RL has been a topic of extensive study over the past years.  ... 
arXiv:2008.07875v1 fatcat:xcot3aktdra6nmq4svxhimx2ju

Adversarial Attacks and Defense in Deep Reinforcement Learning (DRL)-Based Traffic Signal Controllers

Ammar Haydari, Michael Zhang, Chen-Nee Chuah
2021 IEEE Open Journal of Intelligent Transportation Systems  
The results show that the performance of the DRL learning agent decreases in both adversarial attack models with white-box and black-box settings resulting in higher levels of traffic congestion.  ...  In this paper, first, we explore the security vulnerabilities of DRL-based TSCs in the presence of adversarial attacks.  ...  Deep Reinforcement Learning for TSC In this section, we will discuss relevant DRL settings for single-agent and multi-agent settings.  ... 
doi:10.1109/ojits.2021.3118972 fatcat:fbtbpwn4yrgh3k2yzxqvorq3qq

Evaluating the Robustness of Deep Reinforcement Learning for Autonomous and Adversarial Policies in a Multi-agent Urban Driving Environment [article]

Aizaz Sharif, Dusica Marijan
2022 arXiv   pre-print
The results indicate that only some of the deep reinforcement learning algorithms perform consistently better across single and multi-agent scenarios when trained in a multi-agent-only setting.  ...  driving in a single- and multi-agent environment.  ...  Adversarial Reinforcement Learning ARL is a new branch of RL where adversarial algorithms are trained such that they create perturbation attacks against a victim.  ... 
arXiv:2112.11947v2 fatcat:xfmgcgfnrbhe3cimnd4eedoqjq

Evaluating Robustness of Cooperative MARL: A Model-based Approach [article]

Nhan H. Pham, Lam M. Nguyen, Jie Chen, Hoang Thanh Lam, Subhro Das, Tsui-Wei Weng
2022 arXiv   pre-print
In recent years, a proliferation of methods were developed for cooperative multi-agent reinforcement learning (c-MARL).  ...  In addition, we propose the first victim-agent selection strategy which allows us to develop even stronger adversarial attack.  ...  Perhaps unsurprisingly, deep reinforcement learning (DRL) agents are also vulnerable to adversarial attacks, as first shown in Huang et al. [2017] for Atari games DRL agents.  ... 
arXiv:2202.03558v1 fatcat:dkvv3syfjzao3erlbvtghwgj2i

Adversarial attack and defense in reinforcement learning-from AI security view

Tong Chen, Jiqiang Liu, Yingxiao Xiang, Wenjia Niu, Endong Tong, Zhen Han
2019 Cybersecurity  
Hence, in this paper, we give the very first attempt to conduct a comprehensive survey on adversarial attacks in reinforcement learning under AI security.  ...  However, recent studies discover that the interesting attack mode adversarial attack also be effective when targeting neural network policies in the context of reinforcement learning, which has inspired  ...  Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.  ... 
doi:10.1186/s42400-019-0027-x fatcat:nlox7arfojaerietjz5ipskucm

Research and Challenges of Reinforcement Learning in Cyber Defense Decision-Making for Intranet Security

Wenhao Wang, Dingyuanhao Sun, Feng Jiang, Xingguo Chen, Cheng Zhu
2022 Algorithms  
Reinforcement learning has made great breakthroughs in addressing complicated decision-making problems.  ...  Our work provides a systematic view for understanding and solving decision-making problems in the application of reinforcement learning to cyber defense.  ...  Depending on whether the adversary or the environment is considered, single-agent or multi-agent reinforcement learning algorithms can be selected.  ... 
doi:10.3390/a15040134 fatcat:an3gyhnyzve6jj5r74lvqj6eki

Resilient adaptive optimal control of distributed multi-agent systems using reinforcement learning

Rohollah Moghadam, Hamidreza Modares
2018 IET Control Theory & Applications  
This study presents a unified resilient model-free reinforcement learning (RL) based distributed control protocol for leader-follower multi-agent systems.  ...  Although RL has been successfully used to learn optimal control protocols for multi-agent systems, the effects of adversarial inputs are ignored.  ...  Reinforcement learning (RL) [5] [6] [7] [8] [9] [10] , inspired by learning mechanisms observed in mammals, has been successfully used to learn optimal solutions online in single agents for both regulation  ... 
doi:10.1049/iet-cta.2018.0029 fatcat:icfmfh75szgetd3gcf5sbomcyu

A Survey of Adversarial Machine Learning in Cyber Warfare

Vasisht Duddu
2018 Defence Science Journal  
Adversarial machine learning is a fast growing area of research which studies the design of Machine Learning algorithms that are robust in adversarial environments.  ...  in adversarial environments.  ...  in reinforcement learning using transferability across policies to attack the Reinforcement Learning model.  ... 
doi:10.14429/dsj.68.12371 fatcat:vyupcxe6hrhllb4rowequxrf5i
« Previous Showing results 1 — 15 out of 2,164 results