Safety-Constrained Reinforcement Learning for MDPs.
Sebastian JungesNils JansenChristian DehnertUfuk TopcuJoost-Pieter KatoenPublished in: CoRR (2015)
Keyphrases
- reinforcement learning
- markov decision processes
- state space
- optimal policy
- function approximation
- learning algorithm
- control problems
- model based reinforcement learning
- dynamic programming
- average reward
- markov decision process
- finite state
- model free
- temporal difference
- reward function
- reinforcement learning algorithms
- continuous state and action spaces
- machine learning
- state and action spaces
- multi agent
- action sets
- policy evaluation
- markov decision problems
- action space
- function approximators
- planning under uncertainty
- continuous state
- transition model
- single agent
- rl algorithms
- supervised learning
- decision theoretic planning
- average cost
- partially observable
- optimal control
- factored markov decision processes