Title :
An information criterion for optimal neural network selection
Author_Institution :
Orincon Corp., San Diego, CA, USA
fDate :
9/1/1991 12:00:00 AM
Abstract :
The choice of an optimal neural network design for a given problem is addressed. A relationship between optimal network design and statistical model identification is described. A derivative of Akaike´s information criterion (AIC) is given. This modification yields an information statistic which can be used to objectively select a `best´ network for binary classification problems. The technique can be extended to problems with an arbitrary number of classes
Keywords :
identification; information theory; neural nets; optimisation; statistical analysis; Akaike´s information criterion; binary classification; neural network design; optimal selection; statistical model identification; Art; Computational efficiency; Computer networks; Feedforward neural networks; Neural networks; Neurofeedback; Statistics; Training data;
Journal_Title :
Neural Networks, IEEE Transactions on
