|
University of Houston • University of Houston-Clear Lake • ISSO Annual Report Y2005 • 102
The Impact of Chromosome Lineage upon Genetic Program Modeling
One of the challenges in data mining is providing sufficient coverage of the search space in order to produce an acceptable model. Traditionally, Genetic Programs consider all chromosomes within a population for breeding purposes. Considering the expanse of the search space, it is imperative to focus breeding efforts in Genetic Programs in order to attain a better solution in less time. This research examines the lineage of Genetic Programs in order to identify any breeding patterns. Five separate experiments are conducted where chromosomes are grouped into five classes. Lineage patterns are assessed for the best, middle, and worst-class parental chromosomes. Based upon the results, a new Genetic Programming (GP) process is proposed.
The technique utilized in this study solves problems by genetrically breeding a population of individuals, or chromosomes, over a series of generations.
Results
The results are very promising from these experiments. We anticipate that the proposed
approach will lead to significant improvements in genetic programming performance.
Publications
Boetticher, G. and K. Kaminsky. "Assessing the Impact of Chromosome Lineage in
Genetic Program Modeling," Data Mining and Knowledge Discovery J. (Submitted.)
Institute for Space Systems Operations - Y2005 Annual Report
Copyright © 2006
|
