Title :
Parsimonious side propagation
Author :
Bradley, P.S. ; Mangasarian, O.L.
Author_Institution :
Dept. of Comput. Sci., Wisconsin Univ., Madison, WI, USA
Abstract :
A fast parsimonious linear-programming-based algorithm for training neural networks is proposed that suppresses redundant features while using a minimal tra number of hidden units. This is achieved by propagating sideways to newly added hidden units the task of separating successive groups of unclassified points. Computational results show an improvement of 26.53% and 19.76% in tenfold cross-validation test correctness over a parsimonious perceptron on two publicly available datasets
Keywords :
learning (artificial intelligence); linear programming; neural nets; algorithm; cross-validation test correctness; datasets; fast parsimonious linear-programming; neural network training; parsimonious perceptron; parsimonious side propagation; unclassified points separation; Computer networks; Mathematical programming; Neural networks; Particle separators; Testing; Training data; Writing;
Conference_Titel :
Acoustics, Speech and Signal Processing, 1998. Proceedings of the 1998 IEEE International Conference on
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-4428-6
DOI :
10.1109/ICASSP.1998.681829
