Curiosity-driven Exploration in Deep Reinforcement Learning via Bayesian Neural Networks.
Rein HouthooftXi ChenYan DuanJohn SchulmanFilip De TurckPieter AbbeelPublished in: CoRR (2016)
Keyphrases
- reinforcement learning
- neural network
- exploration exploitation
- active exploration
- exploration strategy
- data driven
- function approximators
- belief nets
- pattern recognition
- action selection
- function approximation
- model based reinforcement learning
- learning capabilities
- state space
- back propagation
- bayesian networks
- fuzzy logic
- active learning
- learning algorithm
- autonomous learning
- reinforcement learning algorithms
- machine learning
- multi agent
- artificial neural networks
- markov decision processes
- neural network model
- multilayer perceptron
- robotic control
- dynamic programming
- bayesian estimation
- temporal difference
- bayesian learning
- learning process
- optimal policy
- self organizing maps
- feed forward
- model free
- training process
- fuzzy systems
- posterior distribution
- decision theory
- fault diagnosis
- activation function
- learning classifier systems
- bayesian decision
- multi layer