TY - GEN
T1 - Proposal of new ant system based on consistency and discrepancy of subjective ranking
AU - Uneme, Kotaro
AU - Sakiyama, Tomoko
AU - Arizono, Ikuo
N1 - Funding Information:
This work was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number 17K01266.
PY - 2018
Y1 - 2018
N2 - It is known that the Ant Colony Optimization (ACO) inspired from the collective behavior of real ants, and it is effective to find a better solution for the Traveling Salesman Problem (TSP). Rank based Ant System ASrank has been proposed as a developed version of basic Ant System. In the algorithm of ASrank, each agent in Ant System is ranked from the viewpoint outside the system as to the participation in pheromone update. Then, in spite of the fact that the collective behavior of real ants has inspired in constructing the algorithm of Ant System, ASrank as a developed version includes the viewpoint outside the system that does not exist in the actual ants’ swarm. Furthermore, there is a problem that it tends to be easy to fall into a local solution. In our study, we introduce the behavior observed in real ants’ experiments in order to construct a new algorithm of Ant System. That is, each ant agent in Ant System estimates its own rank by interaction with encountered agents to determine whether it should contribute to pheromone deposition. Therefore, we carried out exploring simulations in several TSP datasets, and we will show some analysis results that indicate the proposed model has superiority than ASrank.
AB - It is known that the Ant Colony Optimization (ACO) inspired from the collective behavior of real ants, and it is effective to find a better solution for the Traveling Salesman Problem (TSP). Rank based Ant System ASrank has been proposed as a developed version of basic Ant System. In the algorithm of ASrank, each agent in Ant System is ranked from the viewpoint outside the system as to the participation in pheromone update. Then, in spite of the fact that the collective behavior of real ants has inspired in constructing the algorithm of Ant System, ASrank as a developed version includes the viewpoint outside the system that does not exist in the actual ants’ swarm. Furthermore, there is a problem that it tends to be easy to fall into a local solution. In our study, we introduce the behavior observed in real ants’ experiments in order to construct a new algorithm of Ant System. That is, each ant agent in Ant System estimates its own rank by interaction with encountered agents to determine whether it should contribute to pheromone deposition. Therefore, we carried out exploring simulations in several TSP datasets, and we will show some analysis results that indicate the proposed model has superiority than ASrank.
UR - http://www.scopus.com/inward/record.url?scp=85062593843&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85062593843&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85062593843
T3 - Lecture Notes in Engineering and Computer Science
BT - Proceedings of the International MultiConference of Engineers and Computer Scientists 2018, IMECS 2018
A2 - Castillo, Oscar
A2 - Feng, David Dagan
A2 - Korsunsky, A.M.
A2 - Douglas, Craig
A2 - Ao, S. I.
PB - Newswood Limited
T2 - 2018 International MultiConference of Engineers and Computer Scientists, IMECS 2018
Y2 - 14 March 2018 through 16 March 2018
ER -