mlsanityData to 7 Oct 2026

How You Act Tells a Lot: Privacy-Leakage Attack on Deep Reinforcement Learning

arXiv 1904.11082 ↗First posted 24 Apr 2019cs.LG stat.MLcited by 29

Not on the map. Why?

Abstract (arXiv)

Machine learning has been widely applied to various applications, some of which involve training with privacy-sensitive data. A modest number of data breaches have been studied, including credit card information in natural language data and identities from face dataset. However, most of these studies focus on supervised learning models. As deep reinforcement learning (DRL) has been deployed in a number of real-world systems, such as indoor robot navigation, whether trained DRL policies can leak private information requires in-depth study. To explore such privacy breaches in general, we mainly propose two methods: environment dynamics search via genetic algorithm and candidate inference based on shadow policies. We conduct extensive experiments to demonstrate such privacy vulnerabilities in DRL under various settings. We leverage the proposed algorithms to infer floor plans from some trained Grid World navigation DRL agents with LiDAR perception. The proposed algorithm can correctly infer most of the floor plans and reaches an average recovery rate of 95.83% using policy gradient trained agents. In addition, we are able to recover the robot configuration in continuous control environments and an autonomous driving simulator with high accuracy. To the best of our knowledge, this is the first work to investigate privacy leakage in DRL settings and we show that DRL-based agents do potentially leak privacy-sensitive information from the trained policies.

Key facts. Cited by 29 papers in the map (data through 7 Oct 2026). First posted on arXiv 24 Apr 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 6; 2022: 4; 2023: 10; 2024: 5; 2025: 2; 2026: 2.

29 papers.

Sort:

By area:

  1. 2023Learning Zero-Shot Cooperation with Humans, Assuming Humans Are Biasedcited by 37
  2. 2021Applications of Multi-Agent Reinforcement Learning in Future Internet: A Comprehensive Surveycited by 34
  3. 2023Reinforcement Unlearningcited by 16
  4. 2022Distributed Differential Privacy in Multi-Armed Banditscited by 14
  5. 2023ORL-AUDITOR: Dataset Auditing in Offline Deep Reinforcement Learningcited by 11
  6. 2022Differentially Private Temporal Difference Learning with Stochastic Nonconvex-Strongly-Concave Optimizationcited by 7
  7. 2022New Challenges in Reinforcement Learning: A Survey of Security and Privacycited by 7
  8. 2024TrajDeleter: Enabling Trajectory Forgetting in Offline Reinforcement Learning Agentscited by 7
  9. 2023adaPARL: Adaptive Privacy-Aware Reinforcement Learning for Sequential-Decision Making Human-in-the-Loop Systemscited by 6
  10. 2021How Private Is Your RL Policy? An Inverse RL Based Analysis Frameworkcited by 5
  11. 2023Multi-Agent Reinforcement Learning: Methods, Applications, Visionary Prospects, and Challengescited by 5
  12. 2023Differentially Private Episodic Reinforcement Learning with Heavy-tailed Rewardscited by 4
17 more
  1. 2023Privacy Risks in Reinforcement Learning for Household Robotscited by 4
  2. 2021Membership Inference Attacks Against Temporally Correlated Data in Deep Reinforcement Learningcited by 3
  3. 2022A Cooperation Graph Approach for Multiagent Sparse Reward Reinforcement Learningcited by 2
  4. 2023Controlling Federated Learning for Covertnesscited by 2
  5. 2025DP-NCB: Privacy Preserving Fair Banditscited by 2
  6. 2021Privacy-Preserving Kickstarting Deep Reinforcement Learning with Privacy-Aware Learnerscited by 1
  7. 2021Reinforcement Learning on Encrypted Datacited by 1
  8. 2021On the Privacy Risks of Deploying Recurrent Neural Networks in Machine Learning Modelscited by 1
  9. 2023Privacy-Engineered Value Decomposition Networks for Cooperative Multi-Agent Reinforcement Learningcited by 1
  10. 2024Differentially Private Deep Model-Based Reinforcement Learningcited by 1
  11. 2024Task-agnostic Decision Transformer for Multi-type Agent Control with Federated Split Trainingcited by 1
  12. 2024Proactive Load-Shaping Strategies with Privacy-Cost Trade-offs in Residential Households based on Deep Reinforcement Learningcited by 1
  13. 2024Privacy Preserving Reinforcement Learning for Population Processescited by 1
  14. 2025PrivORL: Differentially Private Synthetic Dataset for Offline Reinforcement Learningcited by 1
  15. 2026Secure Forgetting: A Framework for Privacy-Driven Unlearning in Large Language Model (LLM)-Based Agentscited by 1
  16. 2026Angel or Demon: Investigating the Plasticity Interventions' Impact on Backdoor Threats in Deep Reinforcement Learningcited by 1
  17. 2023Differentially Private Deep Q-Learning for Pattern Privacy Preservation in MEC Offloadingcited by 0