Beyond the One Step Greedy Approach in Reinforcement Learning.
Yonathan EfroniGal DalalBruno ScherrerShie MannorPublished in: CoRR (2018)
Keyphrases
- control system
- reinforcement learning
- dynamic programming
- post processing
- real time
- greedy algorithm
- optimal policy
- search algorithm
- artificial intelligence
- search space
- feature selection
- state space
- database
- temporal difference
- robotic control
- artificial neural networks
- locally optimal
- batch mode
- learning capabilities
- reinforcement learning algorithms
- preprocessing step
- function approximation
- markov decision processes
- transfer learning
- learning algorithm
- hidden markov models