Observational Overfitting in Reinforcement Learning.
Xingyou SongYiding JiangStephen TuYilun DuBehnam NeyshaburPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- state space
- cross validation
- reinforcement learning algorithms
- function approximation
- decision trees
- markov decision processes
- optimal policy
- model free
- causal inference
- multi agent
- learning algorithm
- neural network
- dynamic programming
- supervised learning
- reinforcement learning methods
- function approximators
- learning problems
- optimal control
- machine learning
- temporal difference
- data sets
- robot control
- real robot
- control problems
- partially observable
- learning capabilities
- reward function
- learning tasks
- monte carlo
- learning process
- active learning
- least squares