The Unintended Consequences of Discount Regularization: Improving Regularization in Certainty Equivalence Reinforcement Learning.
Sarah RathnamSonali ParbhooWeiwei PanSusan A. MurphyFinale Doshi-VelezPublished in: CoRR (2023)
Keyphrases
- reinforcement learning
- regularization parameter
- prior information
- parameter selection
- equivalence relationship
- learning process
- empirical risk minimization
- data dependent
- projection operator
- machine learning
- regularization methods
- regularization method
- blind deconvolution
- temporal difference
- regularization term
- function approximation
- loss function
- state space
- multi agent