Multi-Objective Quality Assessment for EA Parameter Tuning
Evolutionary algorithms are non-deterministic and highly parameterizable optimization methods. Therefore, the setting of parameters greatly influences their performance and methods for parameter tuning became more and more popular in recent years. However, obtained parameter settings are usually valid only for the tackled combination of algorithm, problem, and performance measure. In most investigations concerning EA tuning, only one performance measure is utilized, inherently defining it as ’user preference’. However, users’ preferences may be different. While one user may only look for a single best solution from a couple of runs for a design problem (single-excellent case), another may be interested in a generally stable behavior of the algorithm displayed by a good expected value of multiple runs and a low variance (robust case). An efficient handling of this trade-off is investigated here. In particular, we investigate the possibility to control the behavior of a given algorithm on a given problem between the two stated extremes via changing one or a small number of parameters.
Unable to display preview. Download preview PDF.
- Deb, K., Thiele, L., Laumanns, M., & Zitzler, E. (2002). Scalable multi-objective optimization test problems. In Congress on Evolutionary Computation (CEC 2002) (pp. 825–830). Piscataway, New Jersey: IEEE Press.Google Scholar
- DeJong, K. A. (2006). Evolutionary computation: A unified approach. Cambridge, MA: MIT Press.Google Scholar
- Huang, V. L., Qin, A. K., Deb, K., Zitzler, E., Suganthan, P. N., Liang, J. J., Preuss, M., & Huband, S. (2007). Problem definitions for performance assessment of multi-objective optimization algorithms. Technical report, Nanyang Technological University, Singapore.Google Scholar
- Suganthan, P. N., Hansen, N., Liang, J. J., Deb, K., Chen, Y.-P., Auger, A., & Tiwari, S. (2005). Problem definitions and evaluation criteria for the cec 2005 special session on real-parameter optimization. Technical report, Nanyang Technological University, Singapore, and KanGAL Report 2005005, IIT Kanpur, India.Google Scholar
- Zitzler, E. (1999). Evolutionary algorithms for multiobjective optimization: Methods and applications. PhD thesis, Swiss Federal Institute of Technology (ETH), Zurich, Switzerland.Google Scholar