Trajectory-Aware Eligibility Traces for Off-Policy Reinforcement Learning.
Brett DaleyMartha WhiteChristopher AmatoMarlos C. MachadoPublished in: CoRR (2023)
Keyphrases
- eligibility traces
- reinforcement learning
- reinforcement learning algorithms
- state space
- reinforcement learning methods
- function approximation
- policy evaluation
- model free
- control problems
- learning algorithm
- markov decision processes
- machine learning
- multi agent
- learning speed
- least squares
- policy iteration
- learning process
- markov decision problems
- function approximators
- dynamic programming
- temporal difference
- optimal policy
- cost function
- mobile robot