Pre-training with Synthetic Data Helps Offline Reinforcement Learning.
Zecheng WangChe WangZixuan DongKeith W. RossPublished in: ICLR (2024)
Keyphrases
- synthetic data
- reinforcement learning
- supervised learning
- data sets
- real world
- real time
- real image data
- mri data
- training set
- training examples
- robotic control
- training process
- test set
- training samples
- training phase
- temporal difference
- reinforcement learning algorithms
- online learning
- state space
- active learning
- learning process
- statistically significant
- optimal control
- multi agent
- training algorithm
- multi layer perceptron
- learning algorithm
- machine learning
- neural network