Exploring the Efficacy of Multi-Agent Reinforcement Learning for Autonomous Cyber Defence: A CAGE Challenge 4 Perspective
As cyber threats become increasingly automated and sophisticated, novel solutions must be introduced to improve defence of enterprise networks. Deep Reinforcement Learning (DRL) has demonstrated potential in mitigating these advanced threats. Single DRL Agents have proven utility toward execution of autonomous cyber defence. Despite the success of employing single DRL Agents, this approach presents significant limitations, especially regarding scalability within large enterprise networks. An attractive alternative to the single agent approach is the use of Multi-Agent Reinforcement Learning (MARL). However, developing MARL agents is costly with few options for examining MARL cyber defence techniques against adversarial agents. This paper presents a MARL network security environment, the fourth iteration of the Cyber Autonomy Gym for Experimentation (CAGE) challenges. This challenge was specifically designed to test the efficacy of MARL algorithms in an enterprise network. Our work aims to evaluate the potential of MARL as a robust and scalable solution for autonomous network defence.
Authors
- Craig E. Murphy
- Benjamin Bowman (ORCID: https://orcid.org/0000-0003-2866-1612)
- Ryan Gabrys (ORCID: https://orcid.org/0000-0002-9197-3371)
- Joshua A. Jacobs (ORCID: https://orcid.org/0000-0002-9621-9060)
- Adrian Taylor (ORCID: https://orcid.org/0000-0003-2701-9468)
- Metin Ahiskali
- Ryan Kerr (ORCID: https://orcid.org/0000-0002-3454-7640)
- Dirk Van Bruggen (ORCID: https://orcid.org/0000-0001-9436-5200)
- Étienne Borde (ORCID: https://orcid.org/0000-0002-1418-8193)
- Sunny Fugate (ORCID: https://orcid.org/0000-0002-6733-3219)
- David Bowman (ORCID: https://orcid.org/0000-0002-2064-7624)
- Claire Owens
- Prithviraj Dasgupta (ORCID: https://orcid.org/0000-0003-4601-905X)
- Luis Martínez‐Sobrido (ORCID: https://orcid.org/0000-0001-7084-0804)
- William Valentine
- Ahmad Ridley
- Sarah Farmer (ORCID: https://orcid.org/0000-0003-4752-1041)
- Isaiah J. King (ORCID: https://orcid.org/0000-0003-2866-4135)
- Yiyi Zhang (ORCID: https://orcid.org/0000-0002-2896-1554)
- Hao Huang (ORCID: https://orcid.org/0000-0002-6729-1987)
- Ben Edwards (ORCID: https://orcid.org/0000-0001-6041-5054)
- Li Li (ORCID: https://orcid.org/0000-0003-2990-1614)
- W. T. S. Gould
- Mitchell Kiely
- Christopher R. Moir
- K H Cowan
- Alex Fitts
- Olivia Naish
- Miranda Purchase
- Erich Devendorf
Institutions
- United States Air Force Research Laboratory (US)
- United States Naval Research Laboratory (US)
- Defence Research and Development Canada (CA)
- National Security Agency (US)
- Naval Information Warfare Center Pacific (US)
- University of Canterbury (NZ)
- Cornell University (US)
Publication Details
- Journal
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2025-04-11
- DOI
- https://doi.org/10.1609/aaai.v39i28.35158
- Citations
- 9
- Primary Topic
- Advanced Malware Detection Techniques
- Type
- article
- Field-Weighted Citation Impact
- 4.11