R ˟ R: Rapid eXploration for Reinforcement learning via sampling-based reset distributions and imitation pre-training.
Gagan KhandateTristan Luca SaidiSiqi ShangEric T. ChangYang LiuSeth DennisJohnson AdamsMatei T. CiocarliePublished in: Auton. Robots (2024)
Keyphrases
- reinforcement learning
- exploration strategy
- supervised learning
- exploration exploitation
- function approximation
- active exploration
- action selection
- state space
- training and test data
- probability distribution
- model based reinforcement learning
- learning process
- learning problems
- learning algorithm
- markov decision processes
- monte carlo
- training examples
- reinforcement learning algorithms
- training phase
- model free
- test set
- temporal difference
- training process
- transfer learning
- dynamic programming
- training set
- multi agent
- data sets
- balancing exploration and exploitation
- training algorithm
- motion planning
- random variables
- optimal policy
- semi supervised
- neural network