Regret Minimization for Partially Observable Deep Reinforcement Learning.
Peter H. JinSergey LevineKurt KeutzerPublished in: CoRR (2017)
Keyphrases
- worst case
- partially observable
- regret minimization
- reinforcement learning
- nash equilibrium
- decision problems
- game theoretic
- state space
- markov decision processes
- partial observability
- partially observable domains
- dynamical systems
- np hard
- function approximation
- partially observable environments
- reward function
- markov decision problems
- infinite horizon
- reinforcement learning algorithms
- hidden state
- partial observations
- optimal policy
- multi agent
- action models
- dynamic programming
- game theory
- multi agent learning
- fully observable
- action selection
- state variables
- belief state
- markov decision process
- transfer learning
- supervised learning