Online Target Q-learning with Reverse Experience Replay: Efficiently finding the Optimal Policy for Linear MDPs.
Naman AgarwalSyomantak ChaudhuriPrateek JainDheeraj NagarajPraneeth NetrapalliPublished in: CoRR (2021)
Keyphrases
- optimal policy
- markov decision processes
- reinforcement learning
- state space
- finite horizon
- decision problems
- dynamic programming
- average reward
- infinite horizon
- long run
- policy iteration
- markov decision process
- markov decision problems
- state dependent
- average cost
- finite state
- discount factor
- multistage
- reinforcement learning algorithms
- reward function
- control policies
- initial state
- sufficient conditions
- dynamic programming algorithms
- bayesian reinforcement learning
- learning algorithm
- serial inventory systems
- partially observable markov decision processes
- model free
- lost sales