Title :
SVD-NET: an algorithm that automatically selects network structure
Author :
Psichogios, Dimitris C. ; Ungar, Lyle H.
Author_Institution :
Dept. of Chem. Eng., Pennsylvania Univ., Philadelphia, PA, USA
fDate :
5/1/1994 12:00:00 AM
Abstract :
An algorithm is developed for training feedforward neural networks that uses singular value decomposition (SVD) to identify and eliminate redundant hidden nodes. Minimizing redundancy gives smaller networks, producing models that generalize better and thus eliminate the need of using cross-validation to avoid overfitting. The method is demonstrated by modeling a chemical reactor
Keywords :
feedforward neural nets; redundancy; SVD-NET; chemical reactor; feedforward neural network training; network structure selection; redundancy minimisation; redundant hidden nodes; singular value decomposition; Artificial neural networks; Computational modeling; Learning automata; Machinery; Neural networks; Neurons; Predictive models; Recurrent neural networks; Sun; Training data;
Journal_Title :
Neural Networks, IEEE Transactions on
