Adversarial attack and defense in reinforcement learning-from AI security view
Abstract Reinforcement learning is a core technology for modern artificial intelligence, and it has become a workhorse for AI applications ranging from Atrai Game to Connected and Automated Vehicle System (CAV). Therefore, a reliable RL system is the foundation for the security critical applications...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
SpringerOpen
2019-03-01
|
Series: | Cybersecurity |
Subjects: | |
Online Access: | http://link.springer.com/article/10.1186/s42400-019-0027-x |
id |
doaj-d7e8aae0c60b43ea86a71372a118e3d4 |
---|---|
record_format |
Article |
spelling |
doaj-d7e8aae0c60b43ea86a71372a118e3d42020-11-25T02:24:19ZengSpringerOpenCybersecurity2523-32462019-03-012112210.1186/s42400-019-0027-xAdversarial attack and defense in reinforcement learning-from AI security viewTong Chen0Jiqiang Liu1Yingxiao Xiang2Wenjia Niu3Endong Tong4Zhen Han5Beijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing Jiaotong UniversityBeijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing Jiaotong UniversityBeijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing Jiaotong UniversityBeijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing Jiaotong UniversityBeijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing Jiaotong UniversityBeijing Key Laboratory of Security and Privacy in Intelligent Transportation, Beijing Jiaotong UniversityAbstract Reinforcement learning is a core technology for modern artificial intelligence, and it has become a workhorse for AI applications ranging from Atrai Game to Connected and Automated Vehicle System (CAV). Therefore, a reliable RL system is the foundation for the security critical applications in AI, which has attracted a concern that is more critical than ever. 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 innovative researches in this direction. Hence, in this paper, we give the very first attempt to conduct a comprehensive survey on adversarial attacks in reinforcement learning under AI security. Moreover, we give briefly introduction on the most representative defense technologies against existing adversarial attacks.http://link.springer.com/article/10.1186/s42400-019-0027-xReinforcement learningArtificial intelligenceSecurityAdversarial attackAdversarial exampleDefense |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Tong Chen Jiqiang Liu Yingxiao Xiang Wenjia Niu Endong Tong Zhen Han |
spellingShingle |
Tong Chen Jiqiang Liu Yingxiao Xiang Wenjia Niu Endong Tong Zhen Han Adversarial attack and defense in reinforcement learning-from AI security view Cybersecurity Reinforcement learning Artificial intelligence Security Adversarial attack Adversarial example Defense |
author_facet |
Tong Chen Jiqiang Liu Yingxiao Xiang Wenjia Niu Endong Tong Zhen Han |
author_sort |
Tong Chen |
title |
Adversarial attack and defense in reinforcement learning-from AI security view |
title_short |
Adversarial attack and defense in reinforcement learning-from AI security view |
title_full |
Adversarial attack and defense in reinforcement learning-from AI security view |
title_fullStr |
Adversarial attack and defense in reinforcement learning-from AI security view |
title_full_unstemmed |
Adversarial attack and defense in reinforcement learning-from AI security view |
title_sort |
adversarial attack and defense in reinforcement learning-from ai security view |
publisher |
SpringerOpen |
series |
Cybersecurity |
issn |
2523-3246 |
publishDate |
2019-03-01 |
description |
Abstract Reinforcement learning is a core technology for modern artificial intelligence, and it has become a workhorse for AI applications ranging from Atrai Game to Connected and Automated Vehicle System (CAV). Therefore, a reliable RL system is the foundation for the security critical applications in AI, which has attracted a concern that is more critical than ever. 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 innovative researches in this direction. Hence, in this paper, we give the very first attempt to conduct a comprehensive survey on adversarial attacks in reinforcement learning under AI security. Moreover, we give briefly introduction on the most representative defense technologies against existing adversarial attacks. |
topic |
Reinforcement learning Artificial intelligence Security Adversarial attack Adversarial example Defense |
url |
http://link.springer.com/article/10.1186/s42400-019-0027-x |
work_keys_str_mv |
AT tongchen adversarialattackanddefenseinreinforcementlearningfromaisecurityview AT jiqiangliu adversarialattackanddefenseinreinforcementlearningfromaisecurityview AT yingxiaoxiang adversarialattackanddefenseinreinforcementlearningfromaisecurityview AT wenjianiu adversarialattackanddefenseinreinforcementlearningfromaisecurityview AT endongtong adversarialattackanddefenseinreinforcementlearningfromaisecurityview AT zhenhan adversarialattackanddefenseinreinforcementlearningfromaisecurityview |
_version_ |
1724856259573710848 |