mlsanityData to 14 Oct 2026

CyberWorld: World Models for Sample-Efficient Autonomous Cyber Defense

arXiv 2609.31893 ↗First posted 25 Sep 2026cs.LG cs.AI cs.CRcited by 0

On the map: World models → World model security and GUI agents

also in 1 more
Abstract (arXiv)

Deep reinforcement learning has become a prominent approach to autonomous cyber defense. Existing methods are predominantly model-free and consequently require extensive environment interaction. World models provide an alternative by learning predictive dynamics and optimizing policies through imagined trajectories, yielding substantial gains in sample efficiency in robotics and embodied control. Extending this paradigm to cybersecurity raises a fundamental question: what should constitute the "world" in a cyber world model? We introduce CyberWorld, a Dreamer-style world modeling framework that learns latent cyber dynamics from vector, graph, textual, and multimodal representations of the defended network. Across all four scoreable CyberWheel attack strategies, the graph-based CyberWorld variant exceeds a strategy-agnostic control after 3.6k-15.8k environment steps, compared with millions of steps required by model-free PPO. Across representation choices, graph structure provides greater robustness under topology-dependent attacks, while simpler representations remain competitive in overall performance. Among successful runs, the number of episodes required to reach the control remains approximately constant as network size increases from 15 to 100 hosts. These results establish learned cyber dynamics as a sample-efficient and scalable basis for autonomous defense, and identify world representation as a central design axis for robustness and scalability.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in World model security and GUI agents, one of 11 directions in World models. Builds on 33 of its 49 references that are in the map. First posted on arXiv 25 Sep 2026.

Builds on

33 of 49 references.

References · 33

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2017Graph Attention Networkscited by 8,923
  3. 2019Dream to Control: Learning Behaviors by Latent Imaginationcited by 1,653
  4. 2018Learning Latent Dynamics for Planning from Pixelscited by 1,264
  5. 2023Mastering Diverse Domains through World Modelscited by 1,253
  6. 2021Perceiver: General Perception with Iterative Attentioncited by 1,104
  7. 2020Mastering Atari with Discrete World Modelscited by 1,007
  8. 2024DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planningcited by 386
  9. 2022Transformers are Sample-Efficient World Modelscited by 328
  10. 2024Diffusion for World Modeling: Visual Details Matter in Ataricited by 300
  11. 2025Small Language Models are the Future of Agentic AIcited by 270
  12. 2023Transformer-based World Models Are Happy With 100k Interactionscited by 139
21 more
  1. 2024Teams of LLM Agents can Exploit Zero-Day Vulnerabilitiescited by 95
  2. 2023STORM: Efficient Stochastic Transformer based World Models for Reinforcement Learningcited by 89
  3. 2023Learning to Model the World with Languagecited by 68
  4. 2021Contrasting Centralized and Decentralized Critics in Multi-Agent Reinforcement Learningcited by 56
  5. 2024COMBO: Compositional World Models for Embodied Multi-Agent Cooperationcited by 49
  6. 2022Developing Optimal Causal Cyber-Defence Agents via Cyber Security Simulationcited by 37
  7. 2024GenRL: Multimodal-foundation world models for generalization in embodied agentscited by 31
  8. 2022Scalable Multi-Agent Model-Based Reinforcement Learningcited by 25
  9. 2023NASimEmu: Network Attack Simulator & Emulator for Training Agents Generalizing to Novel Scenarioscited by 20
  10. 2025Learning Transformer-based World Models with Contrastive Predictive Codingcited by 20
  11. 2024Hierarchical Multi-agent Reinforcement Learning for Cyber Network Defensecited by 17
  12. 2025Empowering Multi-Robot Cooperation via Sequential World Modelscited by 9
  13. 2025Accurate and Efficient World Modeling with Masked Latent Transformerscited by 7
  14. 2025Automated Cyber Defense with Generalizable Graph-based Reinforcement Learning Agentscited by 7
  15. 2026SoK: The Pitfalls of Deep Reinforcement Learning for Cybersecuritycited by 4
  16. 2023Deep Reinforcement Learning for Cyber System Defense under Dynamic Adversarial Uncertaintiescited by 3
  17. 2025Learning Optimal Defender Strategies for CAGE-2 using a POMDP Modelcited by 3
  18. 2025Speech World Model: Causal State-Action Planning with Explicit Reasoning for Speechcited by 2
  19. 2026ACDZero: MCTS Agent for Mastering Automated Cyber Defensecited by 2
  20. 2026Learning Ad Hoc Network Dynamics via Graph-Structured World Modelscited by 2
  21. 2026Trident : How to Break Deep Reinforcement Learning Cyber Defenses (Agentic)cited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Distilling Knowledge from Large Language Models into Lightweight Reinforcement Learning Agents for Autonomous Cyber Operations
  2. 2026Towards Hierarchical Cyber Defense with Large Language Models: From Planning to Execution
  3. 2025An Attentive Graph Agent for Topology-Adaptive Cyber Defence
  4. 2024Entity-based Reinforcement Learning for Autonomous Cyber Defence
  5. 2025Towards a Generalisable Cyber Defence Agent for Real-World Computer Networks
  6. 2026Ask the Expert: LLM-Guided Reinforcement Learning for Autonomous Cyber Defense
  7. 2026Beyond Rewards in Reinforcement Learning for Cyber Defence
  8. 2025Large Language Model-Based Reward Design for Deep Reinforcement Learning-Driven Autonomous Cyber Defense