Enhancing supervisory training signals with environmental reinforcement learning using adaptive dynamic programming and artificial neural networks.
Niklas MeltonDonald C. WunschPublished in: ICCI*CC (2016)
Keyphrases
- dynamic programming
- reinforcement learning
- artificial neural networks
- feed forward neural networks
- markov decision processes
- state space
- optimal control
- evolutionary artificial neural networks
- optimal policy
- multi layer perceptron
- signal processing
- neural network
- function approximation
- learning capabilities
- computational intelligence
- benchmark classification problems
- learning algorithm
- supervised learning
- using artificial neural networks
- multi agent
- back propagation
- training examples
- adaptive control
- training phase
- training algorithm
- multilayer perceptron
- training process
- markov decision problems
- machine learning
- neural network model
- genetic algorithm ga
- linear programming
- feed forward
- evolutionary algorithm
- control system
- genetic algorithm
- training set
- modulation scheme
- actor critic
- learning process
- partially observable
- feedforward neural networks
- reinforcement learning algorithms
- stereo matching
- radial basis function
- hidden layer