Deep Predictive Policy Training using Reinforcement Learning.
Ali GhadirzadehAtsuto MakiDanica KragicMårten BjörkmanPublished in: CoRR (2017)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- markov decision process
- approximate dynamic programming
- training set
- function approximation
- state and action spaces
- multi agent
- policy gradient
- control policy
- action space
- reinforcement learning algorithms
- training phase
- neural network
- markov decision processes
- training samples
- actor critic
- supervised learning
- state space
- decision problems
- predictive model
- rl algorithms
- partially observable
- transition model
- training algorithm
- reward function
- optimal control
- planning problems
- sufficient conditions
- online learning
- learning process
- learning algorithm