Relationship between fault tolerance, generalization and the Vapnik-Chervonenkis (VC) dimension of feedforward ANNs.
Dhananjay S. PhatakPublished in: IJCNN (1999)
Keyphrases
- feed forward
- fault tolerance
- vapnik chervonenkis
- vc dimension
- generalization bounds
- artificial neural networks
- learning machines
- back propagation
- upper bound
- sample complexity
- learning theory
- data dependent
- generalization ability
- lower bound
- statistical learning theory
- concept classes
- sample size
- inductive inference
- model selection
- neural network
- uniform convergence
- worst case
- hidden layer
- ranking algorithm
- learning problems
- pac learning
- linear classifiers
- function classes
- euclidean space
- concept class
- bp neural network
- theoretical analysis
- special case
- real valued functions
- compression scheme
- efficient learning
- high dimensional
- support vector
- feature selection
- learning algorithm