Investigating the Properties of Neural Network Representations in Reinforcement Learning.
Han WangErfan MiahiMartha WhiteMarlos C. MachadoZaheer AbbasRaksha KumaraswamyVincent LiuAdam WhitePublished in: CoRR (2022)
Keyphrases
- neural network
- reinforcement learning
- artificial neural networks
- function approximation
- back propagation
- state space
- self organizing maps
- desirable properties
- optimal policy
- feed forward
- recurrent neural networks
- fault diagnosis
- multiple representations
- model free
- network architecture
- training process
- multi layer
- higher level
- knn
- pattern recognition
- learning algorithm