Approximation by neural networks is not continuous.
Paul C. KainenVera KurkováAndrew VogtPublished in: Neurocomputing (1999)
Keyphrases
- neural network
- continuous functions
- pattern recognition
- self organizing maps
- fuzzy logic
- feed forward
- piecewise constant
- multilayer perceptron
- genetic algorithm
- database
- artificial neural networks
- back propagation
- fault diagnosis
- error bounds
- recurrent neural networks
- neural nets
- training algorithm
- approximation schemes
- probabilistic model
- data mining
- feedforward neural networks
- continuous variables
- efficient computation
- activation function
- piecewise linear
- training process
- fuzzy systems
- artificial intelligence
- neural network model
- reinforcement learning