Disentangling Transfer in Continual Reinforcement Learning.
Maciej WolczykMichal ZajacRazvan PascanuLukasz KucinskiPiotr MilosPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- transfer learning
- state space
- function approximation
- knowledge transfer
- optimal policy
- multi agent
- reinforcement learning algorithms
- markov decision processes
- temporal difference
- learning tasks
- control problems
- previously learned
- machine learning
- cross domain
- transferring knowledge
- stochastic approximation
- robotic control
- artificial intelligence
- dynamic programming
- active learning
- learning process
- monte carlo
- case study
- bayesian networks
- genetic algorithm
- relational reinforcement learning
- policy search
- website
- decision trees
- reinforcement learning methods
- robot control
- partially observable
- expert systems
- action selection
- optimal control
- supervised learning
- collaborative filtering
- sufficient conditions