A Scalable Laplace Approximation for Neural Networks.
Hippolyt RitterAleksandar BotevDavid BarberPublished in: ICLR (Poster) (2018)
Keyphrases
- neural network
- expectation propagation
- genetic algorithm
- pattern recognition
- artificial neural networks
- fuzzy logic
- approximation error
- relative error
- highly scalable
- feed forward
- approximation ratio
- gaussian process
- recurrent neural networks
- neural network model
- back propagation
- real time
- database
- neural nets
- multi layer
- error bounds
- activation function
- learning rules
- fault diagnosis
- lightweight
- approximation methods
- polygonal approximation
- np hard