Lower Bounds for Approximation of Some Classes of Lebesgue Measurable Functions by Sigmoidal Neural Networks.
José Luis MontañaCruz E. BorgesPublished in: IWANN (1) (2009)
Keyphrases
- lower bound
- neural network
- activation function
- upper bound
- linear functions
- continuous functions
- vc dimension
- artificial neural networks
- objective function
- feedforward neural networks
- pattern recognition
- branch and bound algorithm
- feed forward neural networks
- branch and bound
- neural nets
- closed form
- back propagation
- np hard
- randomized algorithm
- constant factor
- polynomial approximation
- neural network model
- approximation algorithms
- error bounds
- fuzzy logic
- feed forward
- lower and upper bounds
- class labels
- linear programming relaxation
- optimal solution
- genetic algorithm
- upper and lower bounds
- queueing networks
- worst case
- special case