Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms
Proceedings of the 6th International Conference on Genetic Algorithms
An Experimental Analysis of Robinson-Foulds Distance Matrix Algorithms
ESA '08 Proceedings of the 16th annual European symposium on Algorithms
| Hi-index | 0.00 |
Phylogenetics is concerned with inferring the genealogical relationships between a group of organisms (or taxa), and this relationship is usually expressed as an evolutionary tree. However, inferring the phylogenetic tree is not a trivial task since it is impossible to know the true evolutionary history for a set of organisms. As a result, most phylogenetic analyses rely on effective heuristics for obtaining accurate trees. These heuristics use tree score as a basis for establishing an accurate depiction of evolutionary tree relationships. Relatively little work has been done to analyze the relationship between improving tree scores (fitness) and topological accuracy (distance). In this paper, we present a new fitness-distance correlation coefficient called r fd to quantify the relationship between evolutionary trees. By applying this measure to three biological datasets consisting of 44, 60, and 174 taxa, our results show that improvements in fitness are strongly correlated (r fd 0.8) with topological accuracy to the best-tree-overall . Moreover, we investigated the use of the r fd coefficient if the best overall tree is not available and found similar results. Thus, our results show that r fd is a robust measure with several potential applications such as the development of stopping criteria for phylogenetic search.