Sample complexity of variance-reduced policy gradient: weaker assumptions and lower bounds.
Gabor PaczolayMatteo PapiniAlberto Maria MetelliIstván Á. HarmatiMarcello RestelliPublished in: Mach. Learn. (2024)
Keyphrases
- sample complexity
- lower bound
- policy gradient
- vc dimension
- upper bound
- variance reduction
- sample size
- pac learning
- concept classes
- function approximation
- concept class
- reinforcement learning
- special case
- np hard
- worst case
- theoretical analysis
- learning algorithm
- generalization error
- learning problems
- optimal control
- average case
- supervised learning
- gradient method
- active learning
- objective function
- machine learning
- real valued
- finite state
- training examples
- text categorization
- partially observable markov decision processes
- optimal solution