## The Influence of Run-Time Limits on Choosing Ant System Parameters (2003)

Venue: | In Proc. Genetic and Evolutionary Computation Conference (GECCO 2003 |

Citations: | 7 - 2 self |

### BibTeX

@INPROCEEDINGS{Socha03theinfluence,

author = {Krzysztof Socha},

title = {The Influence of Run-Time Limits on Choosing Ant System Parameters},

booktitle = {In Proc. Genetic and Evolutionary Computation Conference (GECCO 2003},

year = {2003},

pages = {49--60},

publisher = {Springer}

}

### OpenURL

### Abstract

The influence of the allowed running time on the choice of the parameters of an ant system is investigated. It is shown that different parameter values appear to be optimal depending on the algorithm run-time. The performance of the MAX-MIN Ant System (MMAS) on the University Course Timetabling Problem (UCTP) -- a type of constraint satisfaction problem -- is used as an example. The parameters taken into consideration include the type of the local search used, and some typical parameters for MMAS -- the tau_min and rho. It is shown that the optimal parameters depend significantly on the time limits set. Conclusions summarizing the influence of time limits on parameter choice, and possible methods of making the parameter choice more independent from the time limits, are presented.

### Citations

906 | Ant system: optimization by a colony of cooperating agents
- Dorigo, Maniezzo, et al.
- 1996
(Show Context)
Citation Context ...ce, and possible methods of making the parameter choice more independent from the time limits, are presented. 1 Introduction Ant Colony Optimization (ACO) is a metaheuristic proposed by Dorigo et al. =-=[1]-=-. The inspiration of ACO is the foraging behavior of real ants. The basic ingredient of ACO is the use of a probabilistic solution construction mechanism based on stigmergy. ACO has been applied succe... |

86 | A memetic algorithm for university exam timetabling
- Burke, Newall, et al.
- 1996
(Show Context)
Citation Context ... algorithms perform particularly well, when supported by a local search (LS) routine [2, 9, 10]. There were also attempts to design the local search for the particular problem tackled here (the UCTP) =-=[11]-=-. Here, we try to show that although adding an LS to an algorithm improves the results obtained, it is important to carefully choose the type of such LS routine, especially with regard to algorithm ru... |

75 | Ant Colony Optimization for Resource-Constrained Project Scheduling
- Merkle, Middendorf, et al.
- 2002
(Show Context)
Citation Context ...sed on stigmergy. ACO has been applied successfully to numerous combinatorial optimization problems including the traveling salesman problem [2], quadratic assignment problem [3], scheduling problems =-=[4], -=-and others. In this paper we focus on the MAX-MIN Ant System [5] – a version of ACO. ACO, similarly to any other metaheuristic, may be parameterized. This means that in order to be able to deliver o... |

48 | ACO algorithms for the traveling salesman problem,” In: Evolutionary Algorithms in Engineering and Computer
- Stützle, Dorigo
- 1999
(Show Context)
Citation Context ...e use of a probabilistic solution construction mechanism based on stigmergy. ACO has been applied successfully to numerous combinatorial optimization problems including the traveling salesman problem =-=[2], -=-quadratic assignment problem [3], scheduling problems [4], and others. In this paper we focus on the MAX-MIN Ant System [5] – a version of ACO. ACO, similarly to any other metaheuristic, may be para... |

37 | MAX-MIN Ant System and local search for combinatorial optimization problems
- Stützle
- 1999
(Show Context)
Citation Context ...ct value based on quality of the solution. 4 Influence of Local Search It has been shown in the literature that ant algorithms perform particularly well, when supported by a local search (LS) routine =-=[2, 9, 10]-=-. There were also attempts to design the local search for the particular problem tackled here (the UCTP) [11]. Here, we try to show that although adding an LS to an algorithm improves the results obta... |

27 | T.: A comparison of the performance of different metaheuristics on the timetabling problem
- Rossi-Doria, Sampels, et al.
- 2003
(Show Context)
Citation Context ...y, Section 6 summarizes the findings and presents the conclusions drawn. 2 Example Problem The problem used to illustrate the thesis of this paper, is the University Course Timetabling Problem (UCTP) =-=[6,7,8]-=-. It is a type of constraint satisfaction problem. It consists of a set of n events E = {e1,...,en} to be scheduled in a set of i timeslots T = {t1,...,ti}, and a set of j rooms R = {r1,...,rj} in whi... |

23 | Ant algorithms for the university course timetabling problem with regard to the state-of-the-art
- Socha, Sampels, et al.
- 2003
(Show Context)
Citation Context ...y, Section 6 summarizes the findings and presents the conclusions drawn. 2 Example Problem The problem used to illustrate the thesis of this paper, is the University Course Timetabling Problem (UCTP) =-=[6,7,8]-=-. It is a type of constraint satisfaction problem. It consists of a set of n events E = {e1,...,en} to be scheduled in a set of i timeslots T = {t1,...,ti}, and a set of j rooms R = {r1,...,rj} in whi... |

20 | Ant colony optimization: An overview
- Maniezzo, Carbonaro
(Show Context)
Citation Context ...ct value based on quality of the solution. 4 Influence of Local Search It has been shown in the literature that ant algorithms perform particularly well, when supported by a local search (LS) routine =-=[2, 9, 10]-=-. There were also attempts to design the local search for the particular problem tackled here (the UCTP) [11]. Here, we try to show that although adding an LS to an algorithm improves the results obta... |

17 | Improving the ant-system: A detailed report on the MAX -MIN ant system
- Stützle, Hoos
- 1996
(Show Context)
Citation Context ...osed time limits, was a subset of the typical MMAS parameters: evaporation rate ρ and pheromone lower bound τmin. We chose these two parameters among others, as they have been shown in the literatur=-=e [12, 10, 5] to-=- have significant impact on the results obtained by a MAX-MIN Ant System. We generated 110 different sets of these two parameters. We chose the evaporation rate ρ ∈ [0.05,0.50] with the step of 0.0... |

9 |
A MAX -MIN ant system for the university timetabling problem
- Socha, Knowles, et al.
- 2002
(Show Context)
Citation Context ...y, Section 6 summarizes the findings and presents the conclusions drawn. 2 Example Problem The problem used to illustrate the thesis of this paper, is the University Course Timetabling Problem (UCTP) =-=[6,7,8]-=-. It is a type of constraint satisfaction problem. It consists of a set of n events E = {e1,...,en} to be scheduled in a set of i timeslots T = {t1,...,ti}, and a set of j rooms R = {r1,...,rj} in whi... |

5 |
R.F.: A memetic algorithm for university exam timetabling
- Burke, Newall, et al.
- 1996
(Show Context)
Citation Context ...nt algorithms perform particularly well, when supported by a local search (LS) routine [2,9,10]. There were also attempts to design the local search for the particular problem tackled here (the UCTP) =-=[11]-=-. Here, we try to show that although adding an LS to an algorithm improves the results obtained, it is important to carefully choose the type of such LS routine, especially with regard to algorithm ru... |