Name
Abstract | ||
|---|---|---|
We compare genetic algorithms (GA) that solve timetabling problems for universities. Here, we define a timetabling problem as that of finding both a class schedule and room allocation that do not conflict with any constraints. We have already proposed the TGA, which evolves two kinds of genotypes, one related to class scheduling, the other related to room allocation. However, TGA has not been applied to many kinds of problems and it has not been compared with other methods. Here, we present four kinds of genetic algorithms, TGA, STGA, SGA, and SGA2. TGA and STGA use the two kinds of genotypes mentioned above. TGA evolves these genotypes alternately, while STGA evolves those sequentially. SGA and SGA2 are based on the simple GA that uses one genotype. SGA evolves a population to find feasible class scheduling and room allocation, while SGA2 evolves a population to find only a feasible class schedule for which feasible room allocation is possible. These algorithms are implemented and we apply these methods to several timetabling problems generated by an automatic timetabling problem generator. We also implement two methods, SA and TS; experiments by those methods have been performed, where SA is a method based on simulated annealing and TS is a method based on Tabu search. The experimental results show that SGA2 finds feasible solutions for almost all problems, and TGA finds feasible solutions when the average number of subjects per class is high. In addition, STGA finds a feasible solution when the problem has many classes, and SA finds feasible solutions when the number of classes is small and it has many constraints and requests. 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(7): 1–12, 2004; Published online in Wiley InterScience (). DOI 10.1002/scj.10682 © 2004 Wiley Periodicals, Inc. |
Year | DOI | Venue |
|---|---|---|
2004 | 10.1002/scj.v35:7 | Systems and Computers in Japan |
Keywords | Field | DocType |
genetic algorithm | Simulated annealing,Population,Mathematical optimization,Computer science,Scheduling (computing),Artificial intelligence,Timetabling problem,Machine learning,Genetic algorithm,Tabu search | Journal |
Volume | Issue | Citations |
35 | 7 | 12 |
PageRank | References | Authors |
0.85 | 5 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hiroaki Ueda | 1 | 154 | 16.74 |
| Daisuke Ouchi | 2 | 16 | 1.55 |
| Kenichi Takahashi | 3 | 156 | 18.94 |
| Tetsuhiro Miyahara | 4 | 267 | 32.75 |