Discounted Reinforcement Learning Does Not Scale.
Matthew A. F. McDonaldPhilip HingstonPublished in: Comput. Intell. (1997)
Keyphrases
- reinforcement learning
- markov decision processes
- optimal policy
- function approximation
- markov decision process
- dynamic programming
- state space
- average reward
- model free
- reinforcement learning algorithms
- learning process
- action space
- temporal difference learning
- finite state
- infinite horizon
- reinforcement learning methods
- cash flow
- data sets
- robotic control
- temporal difference
- dynamical systems
- scale space
- multi agent systems
- multi agent
- genetic algorithm
- real world