Reinforcement learning with nonstationary reward depending on the episode.
Takeshi ShibuyaSeiji YasunobuPublished in: SMC (2011)
Keyphrases
- non stationary
- reinforcement learning
- state space
- function approximation
- adaptive algorithms
- reinforcement learning algorithms
- markov decision processes
- reward function
- machine learning
- stock price
- random fields
- optimal policy
- eligibility traces
- blind source separation
- concept drift
- autoregressive
- model free
- action selection
- temporal difference
- dynamic programming
- reward shaping
- empirical mode decomposition
- fractional brownian motion
- reinforcement learning methods
- partially observable environments
- state action
- learning algorithm
- average reward
- optimal control
- multi component
- sufficient conditions
- support vector machine
- multi agent