An Enhancement of Generalization Ability in Cascade CorrelationAlgorithm by Avoidance of Overfitting/Overtraining Problem

Authors:
Igor V. Tetko;Alessandro E. P. Villa
Affiliations:
Laboratoire de Neuro-heuristique, Institut de Physiologie, Université de Lausanne, Rue du Bugnon 7, Lausanne, CH–1005, Switzerland;Laboratoire de Neuro-heuristique, Institut de Physiologie, Université de Lausanne, Rue du Bugnon 7, Lausanne, CH–1005, Switzerland
Venue:
Neural Processing Letters
Year:
1997

Citing 7
Cited 2

Neurocomputing

Neurocomputing
The cascade-correlation learning architecture

Advances in neural information processing systems 2
Neural networks and the bias/variance dilemma

Neural Computation
A practical Bayesian framework for backpropagation networks

Neural Computation
Bagging predictors

Machine Learning
Efficient partition of learning data sets for neural network training

Neural Networks
Generalization and regularization in nonlinear learning systems

The handbook of brain theory and neural networks

An Efficient Partition of Training Data Set Improves Speed andAccuracy of Cascade-correlation Algorithm

Neural Processing Letters
A Learning Algorithm for Evolving Cascade Neural Networks

Neural Processing Letters

Quantified Score

Hi-index	0.00

Visualization

Abstract

The current study investigates a method for avoidanceof an overfitting/overtraining problem in ArtificialNeural Network (ANN) based on a combination of twoalgorithms: Early Stopping and Ensemble averaging(ESE). We show that ESE provides an improvement of theprediction ability of ANN trained according to CascadeCorrelation Algorithm. A simple algorithm to estimatethe generalization ability of the method according tothe Leave-One-Out technique is proposed and discussed.In the accompanying paper the problem of optimalselection of training cases is considered foraccelerated learning of the ESE method.