Bridging the Gap Between Value and Policy Based Reinforcement Learning.
Ofir NachumMohammad NorouziKelvin XuDale SchuurmansPublished in: CoRR (2017)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- markov decision process
- action selection
- function approximators
- markov decision processes
- partially observable environments
- reinforcement learning algorithms
- action space
- state and action spaces
- reinforcement learning problems
- state space
- actor critic
- control policy
- policy gradient
- function approximation
- policy iteration
- approximate dynamic programming
- control policies
- model free
- continuous state spaces
- markov decision problems
- dynamic programming
- state action
- temporal difference
- reward function
- machine learning
- decision problems
- partially observable markov decision processes
- partially observable
- rl algorithms
- continuous state
- control problems
- policy evaluation
- approximate policy iteration
- agent learns
- infinite horizon
- transfer learning
- learning process
- partially observable domains
- reinforcement learning methods
- average reward
- model free reinforcement learning
- transition model
- inverse reinforcement learning
- least squares
- multi agent reinforcement learning
- multi agent
- policy making
- learning problems
- gradient method