Asymptotic equipartition property on empirical sequence in reinforcement learning.
Kazunori IwataKazushi IkedaHideaki SakaiPublished in: Neural Networks and Computational Intelligence (2004)
Keyphrases
- reinforcement learning
- hidden state
- asymptotic properties
- machine learning
- state space
- finite sample
- optimal policy
- function approximation
- learning algorithm
- robotic control
- multi agent
- markov decision processes
- desirable properties
- transfer learning
- data sets
- partially observable
- multi agent reinforcement learning
- action selection
- dynamical systems
- policy search
- laplace transform
- model free
- empirical data
- theoretical analysis
- input data
- worst case
- upper bound
- lower bound
- similarity measure
- neural network