Provable Benefits of Representational Transfer in Reinforcement Learning.
Alekh AgarwalYuda SongWen SunKaiwen WangMengdi WangXuezhou ZhangPublished in: COLT (2023)
Keyphrases
- reinforcement learning
- transfer learning
- knowledge transfer
- previously learned
- function approximation
- machine learning
- reinforcement learning algorithms
- learning tasks
- transferring knowledge
- robotic control
- real robot
- control problems
- temporal difference
- state space
- dynamic programming
- search algorithm
- learning algorithm
- real world
- optimal policy
- data sets
- supervised learning
- model free
- action selection
- optimal solution
- multi agent
- control policy
- temporal difference learning
- reinforcement learning methods
- genetic algorithm