Finite Time Guarantees for Continuous State MDPs with Generative Model.
Hiteshi SharmaRahul JainPublished in: CDC (2020)
Keyphrases
- generative model
- continuous state
- reinforcement learning
- continuous state and action spaces
- policy search
- markov decision processes
- action space
- finite state
- continuous state spaces
- partially observable markov decision processes
- robot navigation
- probabilistic model
- state space
- control policies
- state dependent
- optimal policy
- em algorithm
- conditional random fields
- semi supervised
- reward function
- planning problems
- prior knowledge
- initial state
- state action
- function approximation
- real valued
- dynamical systems
- multi agent
- learning algorithm
- partially observable
- topic models
- markov decision process
- finite horizon
- markov chain
- dynamic programming
- markov decision problems
- autonomous robots
- function approximators
- action selection
- model free
- infinite horizon
- model checking
- bayesian networks