Limitations on approximation by deep and shallow neural networks.
Guergana PetrovaPrzemyslaw WojtaszczykPublished in: CoRR (2022)
Keyphrases
- neural network
- pattern recognition
- back propagation
- artificial neural networks
- fuzzy logic
- question answering
- error bounds
- approximation methods
- genetic algorithm
- approximation error
- natural language processing
- neural network model
- wall street journal
- deep learning
- multilayer perceptron
- closed form
- radial basis function
- approximation algorithms
- fuzzy systems
- training algorithm
- fuzzy neural network
- queueing networks
- rule extraction
- competitive learning
- information extraction
- support vector machine
- continuous functions
- data sets
- deep knowledge
- deep architectures
- real time