Deep auto-encoder neural networks in reinforcement learning.
Sascha LangeMartin A. RiedmillerPublished in: IJCNN (2010)
Keyphrases
- neural network
- reinforcement learning
- function approximators
- pattern recognition
- reinforcement learning algorithms
- low complexity
- bit rate
- rate distortion
- learning classifier systems
- function approximation
- artificial neural networks
- learning capabilities
- single layer
- state space
- fuzzy logic
- model free
- recurrent neural networks
- video compression
- genetic algorithm
- neural nets
- multilayer perceptron
- back propagation
- neural network model
- multi agent
- distributed video coding
- temporal difference
- action selection
- deep learning
- machine learning
- power reduction
- multi layer
- feed forward
- self organizing maps
- multi layer perceptron
- activation function
- training process
- motion compensated
- reward function
- optimal control
- video codec
- radial basis function
- fault diagnosis
- dynamic programming
- video sequences
- robotic control