Details for NCL-TR-2008004
PropertyValue
NameNCL-TR-2008004
Description
Sensible Linkage, Effective Distributions, and Model Pruning in Estimation of Distribution Algorithms
Chuang, Chung-Yao
Abstract: Estimation of distribution algorithms (EDAs) are a class of evolutionary algorithms that replace the traditional variation operators, such as mutation and crossover, by building a probabilistic model on promising solutions and sampling the built model to generate new candidate solutions. Using probabilistic models for exploration enables these methods to use advanced techniques of statistics and machine learning for automatic discovery of problem structures. However, in some situations, complete and accurate identification of all problem structures by probabilistic modeling is not possible because of certain inherent properties of the given problem. In this work, we illustrate one possible cause of such situations with problems composed of structures of unequal fitness contributions. Based on the illustrative example, a notion is introduced that the estimated probabilistic models should be inspected to reveal the effective search directions, and we propose a general approach which utilizes a reserved set of solutions to examine the built model for likely inaccurate fragments. Furthermore, the proposed approach is implemented in the extended compact genetic algorithm (ECGA) and experimented on several sets of problems with different scaling difficulties. The results indicate that the proposed method can significantly assist ECGA to handle problems comprising structures of disparate fitness contributions and therefore may potentially help EDAs in general to overcome those situations in which the entire structure of the problem cannot be recognized properly due to the temporal delay of emergence of some promising partial solutions.
FilenameNCL-TR-2008004.pdf
Filesize646.72 kB
Filetypepdf (Mime Type: application/pdf)
Creatorypchen
Created On: 08/13/2008 12:12
ViewersEverybody
Maintained byEditor
Hits2763 Hits
Last updated on 12/09/2010 14:04
Homepage