Closed Form Variational Objectives For Bayesian Neural Networks with a Single Hidden Layer.
Martin JankowiakPublished in: CoRR (2018)
Keyphrases
- closed form
- single hidden layer
- neural network
- feedforward neural networks
- extreme learning machine
- activation function
- hidden layer
- feed forward neural networks
- hyperparameters
- back propagation
- marginal likelihood
- gaussian processes
- artificial neural networks
- hidden neurons
- neural network model
- genetic algorithm
- multilayer perceptron
- bayesian networks
- posterior distribution
- network architecture
- neural nets
- recurrent neural networks
- feed forward
- fault diagnosis
- error function
- maximum likelihood
- bayesian inference
- support vector regression
- training process
- radial basis function
- optical flow
- feature vectors