DocumentCode :
285317
Title :
Neural network architecture for linear programming
Author :
Caudell, Thomas P. ; Zikan, Karel
Author_Institution :
Boeing Computer Services, Seattle, WA, USA
Volume :
3
fYear :
1992
fDate :
7-11 Jun 1992
Firstpage :
91
Abstract :
A neural network architecture, called LP-Net, is introduced that rapidly solves general linear programming problems. Mathematically, the approach is based on the logarithmic barrier function approach to linear programming. The neural network simulates the barrier method´s first-order dynamic system. The authors briefly outline the logarithmic barrier technique, present the neural network architecture, and give the set of differential equations that describes the network dynamics. The convergence properties of this neural network makes it ideal for analog hardware implementation
Keywords :
linear programming; neural nets; LP-Net; first-order dynamic system; linear programming; logarithmic barrier function approach; neural network architecture; Analog computers; Computational modeling; Computer architecture; Differential equations; Dynamic programming; Linear programming; Neural network hardware; Neural networks; Optimization methods; Transportation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Networks, 1992. IJCNN., International Joint Conference on
Conference_Location :
Baltimore, MD
Print_ISBN :
0-7803-0559-0
Type :
conf
DOI :
10.1109/IJCNN.1992.227185
Filename :
227185
Link To Document :
بازگشت