Sampling Attacks on Meta Reinforcement Learning: A Minimax Formulation and Complexity Analysis.
Tao LiHaozhe LeiQuanyan ZhuPublished in: CoRR (2022)
Keyphrases
- complexity analysis
- reinforcement learning
- theoretical analysis
- function approximation
- meta level
- lower bound
- first order logic
- computational complexity
- state space
- sampling strategy
- markov decision processes
- optimal policy
- reinforcement learning algorithms
- model free
- watermarking scheme
- sampling methods
- temporal difference
- machine learning
- countermeasures
- worst case
- random sampling
- optimal control
- learning process
- learning algorithm
- neural network
- monte carlo
- security threats
- security risks
- malicious attacks
- evaluation function
- supervised learning
- search algorithm