Reinforcement Learning Methods to Handle Actions with Differing Costs in MDPs.
Takahisa IshiguroTohgoroh MatsuiNobuhiro InuzukaKoichi WadaPublished in: KES (2003)
Keyphrases
- reinforcement learning methods
- action space
- markov decision processes
- reinforcement learning
- reinforcement learning algorithms
- state space
- reward function
- partially observable
- average cost
- reinforcement learning problems
- real valued
- dynamical systems
- action selection
- learning agent
- markov decision problems
- stochastic processes
- optimal policy
- control problems
- initial state
- finite state
- dynamic programming
- markov decision process
- policy iteration
- learning algorithm
- control strategies
- model construction
- decision theoretic
- average reward
- machine learning
- multi agent
- search algorithm
- function approximators
- learning process
- mobile robot
- fuzzy logic
- markov chain
- transfer learning
- planning problems
- function approximation
- optimal control
- multiple agents
- model free