Exploring parameter space in reinforcement learning.
Thomas RückstießFrank SehnkeTom SchaulDaan WierstraYi SunJürgen SchmidhuberPublished in: Paladyn J. Behav. Robotics (2010)
Keyphrases
- parameter space
- reinforcement learning
- image space
- multivariate data
- starting points
- high dimensional
- function approximation
- reinforcement learning algorithms
- parameter values
- model free
- likelihood function
- state space
- markov decision processes
- temporal difference
- robotic control
- learning problems
- multi agent
- learning algorithm
- action selection
- search space
- lie group
- linear transformations
- dynamic programming
- real robot
- neural network