Adaptive Reward-Poisoning Attacks against Reinforcement Learning.
Xuezhou ZhangYuzhe MaAdish SinglaXiaojin ZhuPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- function approximation
- adaptive control
- state space
- machine learning
- partially observable environments
- temporal difference
- markov decision processes
- optimal policy
- model free
- learning process
- average reward
- malicious attacks
- action selection
- social networks
- learning capabilities
- policy gradient
- denial of service attacks
- actor critic
- learning algorithm