Locally Differentially Private Reinforcement Learning for Linear Mixture Markov Decision Processes.
Chonghua LiaoJiafan HeQuanquan GuPublished in: ACML (2022)
Keyphrases
- markov decision processes
- differentially private
- reinforcement learning
- reinforcement learning algorithms
- optimal policy
- state space
- policy iteration
- finite state
- differential privacy
- dynamic programming
- state and action spaces
- average reward
- model based reinforcement learning
- function approximation
- partially observable
- action space
- transition matrices
- average cost
- action sets
- markov decision process
- reward function
- decision theoretic planning
- machine learning
- function approximators
- markov decision problems
- state abstraction
- model free
- actor critic
- transfer learning
- discounted reward
- least squares