@article{MDD6D35B9,
title = "Trends in Autonomous Course of Action for Cyber Attacks Using Reinforcement Learning",
journal = "The Journal of Korean Institute of Communications and Information Sciences",
year = "2022",
issn = "1226-4717",
doi = "10.7840/kics.2022.47.11.1776",
author = "Seok Bin Son, Haemin Lee, Soohyun Park, Dong Hwa Kim, Joongheon Kim",
keywords = "Cyber Attacks, Attack COA, Reinforcement Learning, Penetration Testing",
abstract = "As the network is getting large and complex with massively connected endpoints, the security threats from malicious adversaries have continuously increased. Various testing methods have been developed for the prior security evaluation of a target network to derive a series of strategic attack actions that we call Course of Action (COA) by emulating real active adversaries. However, current attack COA techniques require extensive human effort and cost, especially for large and complex networks, and the need for autonomous attack COA on the uncertain state of the target network has emerged. In this context, the reinforcement learning-based approach is regarded as a promising solution. This paper introduces an overview of the state-of-the-art research in RL-based attack COA strategies with the remaining limitations."
}