Fully Bayesian Recurrent Neural Networks for Safe Reinforcement Learning.
Matt BenatanEdward O. Pyzer-KnappPublished in: CoRR (2019)
Keyphrases
- recurrent neural networks
- fully bayesian
- reinforcement learning
- maximum likelihood
- feed forward
- markov chain monte carlo
- generative model
- neural network
- recurrent networks
- gaussian process
- reservoir computing
- posterior distribution
- echo state networks
- parameter estimation
- artificial neural networks
- state space
- dynamic programming
- cascade correlation
- machine learning
- graphical models
- nonlinear dynamic systems
- support vector