Optimizing for the Future in Non-Stationary MDPs.
Yash ChandakGeorgios TheocharousShiv ShankarMartha WhiteSridhar MahadevanPhilip S. ThomasPublished in: CoRR (2020)
Keyphrases
- non stationary
- markov decision processes
- reinforcement learning
- finite horizon
- autoregressive
- adaptive algorithms
- temporal evolution
- stock price
- long term
- random fields
- blind source separation
- state space
- empirical mode decomposition
- change point detection
- factored mdps
- concept drift
- gaussian mixture model
- white noise
- dynamic programming