On the Expressivity of Neural Networks for Deep Reinforcement Learning.
Kefan DongYuping LuoTianhe YuChelsea FinnTengyu MaPublished in: ICML (2020)
Keyphrases
- neural network
- reinforcement learning
- function approximation
- function approximators
- artificial neural networks
- pattern recognition
- markov decision processes
- learning capabilities
- state space
- multilayer perceptron
- deep learning
- reinforcement learning algorithms
- learning classifier systems
- neural network model
- machine learning
- genetic algorithm
- learning problems
- rule extraction
- learning algorithm
- multi agent
- model free
- fuzzy logic
- network architecture
- fault diagnosis
- multi layer
- training process
- optimal control
- back propagation
- neural nets
- action selection
- temporal difference
- recurrent neural networks
- data sets
- radial basis function
- action space
- self organizing maps
- stochastic approximation