Multi-Agent Reinforcement Learning for Cybersecurity: Classification and survey

Salvo Finistrella, Stefano Mariani, Franco Zambonelli

Abstract: Highlights•A classification of reinforcement learning methods applied to cybersecurity.•Novel framework linking MARL Frameworks to specific network configurations for defense.•Discussion of adversarial multi-agent strategies to address evolving and dynamic cyber threats.•Review of MARL applications in SDN environments for enhanced scalability and security.•Panoramic view of multi-agent strategies for network-wide defense against advanced cyber attacks.