Abstract
We propose an industrial technological solution for the traveling salesman problem (TSP) by using the approximated cost matrix and an accelerated evolutionary computation (EC) algorithm. The cost matrix used by theoretical research on TSP mostly is the Euclidean distance between cities, which is not proper to the real condition in the industrial product's application. In this paper, we propose an approximation approach on cost matrix based on the geographic information data, so that it approaches to the actual cost matrix. Slow convergence is the main issue of EC. We propose an accelerating EC convergence approach by Lagrange interpolation method to approximate the EC search space landscape, and do a local search near the related best individuals' region. The experimental result shows that the EC convergence is accelerated, and this acceleration approach is also used in an actual TSP application in a vehicle navigation system, in which the product performance is improved by the accelerated EC approach with the approximated cost matrix.
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011 |
| Pages | 39-44 |
| Number of pages | 6 |
| DOIs | |
| Publication status | Published - Dec 21 2011 |
| Event | 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011 - Dalian, China Duration: Oct 14 2011 → Oct 16 2011 |
Publication series
| Name | Proceedings of the 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011 |
|---|
Other
| Other | 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011 |
|---|---|
| Country | China |
| City | Dalian |
| Period | 10/14/11 → 10/16/11 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Computer Vision and Pattern Recognition
- Software
Cite this
A novel traveling salesman problem solution by accelerated evolutionary computation with approximated cost matrix in an industrial application. / Pei, Yan; Takagi, Hideyuki.
Proceedings of the 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011. 2011. p. 39-44 6089092 (Proceedings of the 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - A novel traveling salesman problem solution by accelerated evolutionary computation with approximated cost matrix in an industrial application
AU - Pei, Yan
AU - Takagi, Hideyuki
PY - 2011/12/21
Y1 - 2011/12/21
N2 - We propose an industrial technological solution for the traveling salesman problem (TSP) by using the approximated cost matrix and an accelerated evolutionary computation (EC) algorithm. The cost matrix used by theoretical research on TSP mostly is the Euclidean distance between cities, which is not proper to the real condition in the industrial product's application. In this paper, we propose an approximation approach on cost matrix based on the geographic information data, so that it approaches to the actual cost matrix. Slow convergence is the main issue of EC. We propose an accelerating EC convergence approach by Lagrange interpolation method to approximate the EC search space landscape, and do a local search near the related best individuals' region. The experimental result shows that the EC convergence is accelerated, and this acceleration approach is also used in an actual TSP application in a vehicle navigation system, in which the product performance is improved by the accelerated EC approach with the approximated cost matrix.
AB - We propose an industrial technological solution for the traveling salesman problem (TSP) by using the approximated cost matrix and an accelerated evolutionary computation (EC) algorithm. The cost matrix used by theoretical research on TSP mostly is the Euclidean distance between cities, which is not proper to the real condition in the industrial product's application. In this paper, we propose an approximation approach on cost matrix based on the geographic information data, so that it approaches to the actual cost matrix. Slow convergence is the main issue of EC. We propose an accelerating EC convergence approach by Lagrange interpolation method to approximate the EC search space landscape, and do a local search near the related best individuals' region. The experimental result shows that the EC convergence is accelerated, and this acceleration approach is also used in an actual TSP application in a vehicle navigation system, in which the product performance is improved by the accelerated EC approach with the approximated cost matrix.
UR - http://www.scopus.com/inward/record.url?scp=83655202620&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=83655202620&partnerID=8YFLogxK
U2 - 10.1109/SoCPaR.2011.6089092
DO - 10.1109/SoCPaR.2011.6089092
M3 - Conference contribution
AN - SCOPUS:83655202620
SN - 9781457711947
T3 - Proceedings of the 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011
SP - 39
EP - 44
BT - Proceedings of the 2011 International Conference of Soft Computing and Pattern Recognition, SoCPaR 2011
ER -