Title :
Parameters rules of the Hopfield/Tank model on solving TSP
Author :
Rong, Liu ; Ze-min, Liu
Author_Institution :
Dept. of Radio Eng., Beijing Univ. of Posts & Telecommun., China
Abstract :
The Hopfield/Tank model has been used to solve large-scale optimization problems, such as the traveling salesman problem (TSP). The model is very sensitive to the parameters (A,B,C,D, and Ne). Following the subspace analysis of the connection matrix in which D=0, the influence of parameter D is investigated with some theorems to be proved. The dynamics of the model are analyzed. Based on these analyses, the parameter rules are established. Some simulation results of 10 city, 30 city and 50 city TSPs are also included. It is shown that if the parameters are selected based on these rules, not only the valid but also the sub-optimal solutions can be found by the Hopfield/Tank model
Keywords :
Hopfield neural nets; operations research; optimisation; connection matrix; large-scale optimization problems; parameter rules; subspace analysis; traveling salesman problem; Analytical models; Cities and towns; Joining processes; Large-scale systems; Runtime; Simulated annealing; Temperature sensors; Traveling salesman problems;
Conference_Titel :
Neural Networks, 1992. IJCNN., International Joint Conference on
Conference_Location :
Baltimore, MD
Print_ISBN :
0-7803-0559-0
DOI :
10.1109/IJCNN.1992.227296
