Two parameter update schemes for recurrent reinforcement learning.
Jin ZhangDietmar MaringerPublished in: IEEE Congress on Evolutionary Computation (2014)
Keyphrases
- reinforcement learning
- function approximation
- recurrent neural networks
- machine learning
- reinforcement learning algorithms
- input parameters
- feed forward
- robotic control
- learning algorithm
- parameter values
- multi agent
- supervised learning
- neural network
- transfer learning
- dynamic programming
- learning process
- parameter space
- parameter settings
- database systems
- parameter tuning
- model free
- action selection
- optimal policy
- update rule