LCRL: Certified Policy Synthesis via Logically-Constrained Reinforcement Learning.
Hosein HasanbeigDaniel KroeningAlessandro AbatePublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- markov decision process
- state space
- partially observable
- reinforcement learning problems
- markov decision processes
- control policies
- partially observable environments
- reinforcement learning algorithms
- control policy
- function approximation
- action space
- actor critic
- policy iteration
- function approximators
- policy gradient
- approximate dynamic programming
- policy evaluation
- markov decision problems
- infinite horizon
- reward function
- state and action spaces
- state dependent
- continuous state spaces
- state action
- exploration exploitation tradeoff
- partially observable domains
- partially observable markov decision processes
- control problems
- rl algorithms
- decision problems
- model free
- model free reinforcement learning
- temporal difference learning
- inverse reinforcement learning
- long run
- approximate policy iteration
- temporal difference
- dynamic programming
- agent receives
- agent learns
- learning agent
- reinforcement learning methods
- program synthesis