Compliant skills acquisition and multi-optima policy search with EM-based reinforcement learning.
Sylvain CalinonPetar KormushevDarwin G. CaldwellPublished in: Robotics Auton. Syst. (2013)
Keyphrases
- policy search
- reinforcement learning
- continuous state
- reinforcement learning algorithms
- dynamic programming
- expectation maximization
- policy gradient
- partially observable markov decision processes
- reward function
- function approximation
- em algorithm
- markov decision processes
- learning problems
- continuous action
- evolutionary algorithm
- optimal solution
- maximum likelihood
- partially observable
- neural network
- generative model
- control strategies
- temporal difference
- state space
- probabilistic model
- learning agent
- function approximators
- hidden state
- learning algorithm
- machine learning