Reinforcement Learning with Temporal Logic Constraints for Partially-Observable Markov Decision Processes.
Yu WangAlper Kamil BozkurtMiroslav PajicPublished in: CoRR (2021)
Keyphrases
- temporal logic
- partially observable markov decision processes
- reinforcement learning
- model checking
- finite state
- continuous state
- partially observable domains
- automata theoretic
- optimal policy
- dynamic constraints
- modal logic
- planning under uncertainty
- markov decision processes
- partial observability
- state space
- partially observable environments
- function approximation
- belief state
- partially observable
- belief revision
- hidden state
- verification method
- dynamic programming
- dynamical systems
- linear temporal logic
- learning algorithm
- belief space
- model free
- fully observable
- decision problems
- planning problems
- reinforcement learning algorithms
- partially observable markov decision process
- policy gradient
- machine learning
- dec pomdps
- initial state
- infinite horizon
- bayesian networks