mlsanityData to 7 Oct 2026

Efficient Reward Poisoning Attacks on Online Deep Reinforcement Learning

arXiv 2205.14842 ↗First posted 30 May 2022cs.LG cs.CRcited by 9

Not on the map. Why?

Abstract (arXiv)

We study reward poisoning attacks on online deep reinforcement learning (DRL), where the attacker is oblivious to the learning algorithm used by the agent and the dynamics of the environment. We demonstrate the intrinsic vulnerability of state-of-the-art DRL algorithms by designing a general, black-box reward poisoning framework called adversarial MDP attacks. We instantiate our framework to construct two new attacks which only corrupt the rewards for a small fraction of the total training timesteps and make the agent learn a low-performing policy. We provide a theoretical analysis of the efficiency of our attack and perform an extensive empirical evaluation. Our results show that our attacks efficiently poison agents learning in several popular classical control and MuJoCo environments with a variety of state-of-the-art DRL algorithms, such as DQN, PPO, SAC, etc.

Key facts. Cited by 9 papers in the map (data through 7 Oct 2026). Builds on 30 of its 42 references that are in the map. First posted on arXiv 30 May 2022.

Builds on

30 of 42 references.

References · 30

  1. 2017Deep reinforcement learning from human preferencescited by 3,846
  2. 2015Deep Reinforcement Learning with Double Q-learningcited by 1,704
  3. 2015Dueling Network Architectures for Deep Reinforcement Learningcited by 769
  4. 2017Deep Reinforcement Learning that Matterscited by 727
  5. 2018Is Q-learning Provably Efficient?cited by 485
  6. 2017Adversarial Attacks on Neural Network Policiescited by 314
  7. 2021Reinforcement learning based recommender systems: A surveycited by 187
  8. 2017Tactics of Adversarial Attack on Deep Reinforcement Learning Agentscited by 163
  9. 2021Robust Reinforcement Learning on State Observations with Learned Optimal Adversarycited by 133
  10. 2020Adversarial Machine Learning -- Industry Perspectivescited by 121
  11. 2018Stochastic bandits robust to adversarial corruptionscited by 118
  12. 2017Delving into adversarial attacks on deep policiescited by 104
18 more
  1. 2017Vulnerability of Deep Reinforcement Learning to Policy Induction Attackscited by 91
  2. 2019Better Algorithms for Stochastic Bandits with Adversarial Corruptionscited by 91
  3. 2020Adaptive Reward-Poisoning Attacks against Reinforcement Learningcited by 86
  4. 2020Policy Teaching via Environment Poisoning: Training-time Adversarial Attacks against Reinforcement Learningcited by 81
  5. 2019Corruption-robust exploration in episodic reinforcement learningcited by 72
  6. 2019Policy Poisoning in Batch Reinforcement Learning and Controlcited by 70
  7. 2018Adversarial Attacks on Stochastic Banditscited by 56
  8. 2019Q-learning with UCB Exploration is Sample Efficient for Infinite-Horizon MDPcited by 53
  9. 2019Data Poisoning Attacks on Stochastic Banditscited by 52
  10. 2020Vulnerability-Aware Poisoning Mechanism for Online RL with Unknown Dynamicscited by 38
  11. 2017Whatever Does Not Kill Deep Reinforcement Learning, Makes It Strongercited by 35
  12. 2021Robust Policy Gradient against Strong Data Corruptioncited by 30
  13. 2019TrojDRL: Trojan Attacks on Deep Reinforcement Learning Agentscited by 29
  14. 2021Provably Efficient Black-Box Action Poisoning Attacks Against Reinforcement Learningcited by 29
  15. 2021Reward Poisoning in Reinforcement Learning: Attacks Against Unknown Learners in Unknown Environmentscited by 26
  16. 2021Improved Corruption Robust Algorithms for Episodic Reinforcement Learningcited by 20
  17. 2019Snooping Attacks on Deep Reinforcement Learningcited by 17
  18. 2022Admissible Policy Teaching through Reward Designcited by 12

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 2; 2024: 2; 2025: 2; 2026: 3.

9 papers.

  1. 2025Wolfpack Adversarial Attack for Robust Multi-Agent Reinforcement Learningcited by 6
  2. 2023Black-Box Targeted Reward Poisoning Attack Against Online Deep Reinforcement Learningcited by 5
  3. 2024Universal Black-Box Reward Poisoning Attack against Offline Reinforcement Learningcited by 4
  4. 2023A Reminder of its Brittleness: Language Reward Shaping May Hinder Learning for Instruction Following Agentscited by 2
  5. 2024Robust Deep Reinforcement Learning against Adversarial Behavior Manipulationcited by 2
  6. 2025Backdoors in DRL: Four Environments Focusing on In-distribution Triggerscited by 1
  7. 2026When Can You Poison Rewards? A Tight Characterization of Reward Poisoning in Linear MDPscited by 0
  8. 2026Efficient Preference Poisoning Attack on Offline RLHFcited by 0
  9. 2026When Critics Disagree: Adaptive Reward Poisoning Attacks in RIS-Aided Wireless Control Systemcited by 0