Self-Healing Mechanisms for Autonomous Cybersecurity Systems Using Reinforcement Learning
Stout in Computer Science and Technology Studies · 30.12.2025 · 1 (1) · 21-30
DOI: 10.61424/0yb2dw48 ↗Kurz gefasst
Fachartikel in der Zeitschrift Stout in Computer Science and Technology Studies, online erschienen am 30.12.2025. Verfasst von Michael David Johnson und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The increasing sophistication of cyber threats has driven the need for intelligent, adaptive cybersecurity systems capable of autonomous operation. Conventional security approaches, such as signature-based and rule-based detection mechanisms, often struggle to identify and respond effectively to evolving and previously unseen attack patterns. This study presents a reinforcement learning (RL)-based self-healing framework that enhances autonomous cybersecurity by improving threat detection, accelerating response and recovery, and increasing overall system resilience. The proposed framework employs reinforcement learning agents that continuously interact with the operating environment to learn and optimize defense strategies. Through ongoing adaptation, the system effectively detects emerging threats and autonomously initiates appropriate mitigation and recovery actions. Comparative performance analysis demonstrates that the RL-based framework consistently outperforms traditional security approaches across multiple cybersecurity metrics. Specifically, the proposed model achieves a threat detection accuracy of approximately 93%, exceeding the performance of conventional signature-based and anomaly-based methods, thereby demonstrating superior adaptability to previously unseen attacks. The framework also significantly reduces post-attack recovery time, achieving system restoration in approximately 30 seconds compared with substantially longer recovery periods required by manual and rule-based approaches. This rapid self-healing capability minimizes service disruption, limits the propagation of attacks, and enhances operational continuity, particularly in critical infrastructure environments. Furthermore, experimental evaluation indicates notable improvements in system resilience, including higher uptime, more effective threat mitigation, and improved response efficiency. By continuously refining its defensive policies through reinforcement learning, the proposed framework provides a robust, adaptive, and resilient cybersecurity solution capable of addressing the challenges posed by dynamic and sophisticated cyber threats.
Zitieren
APA
Johnson, M. D., & Robert, W. (2025). Self-Healing Mechanisms for Autonomous Cybersecurity Systems Using Reinforcement Learning. Stout in Computer Science and Technology Studies, 1(1), 21-30. https://doi.org/10.61424/0yb2dw48
Vancouver
Johnson MD, Robert W. Self-Healing Mechanisms for Autonomous Cybersecurity Systems Using Reinforcement Learning. Stout in Computer Science and Technology Studies. 2025;1(1):21-30. doi:10.61424/0yb2dw48
BibTeX
@article{Johnson2025,
author = {Johnson, Michael David and Robert, William},
title = {Self-Healing Mechanisms for Autonomous Cybersecurity Systems Using Reinforcement Learning},
journal = {Stout in Computer Science and Technology Studies},
year = {2025},
volume = {1},
number = {1},
pages = {21--30},
doi = {10.61424/0yb2dw48}
}
EUVIMED