Abstract
Experimental results are presented of a study on numerical taxonomy on data. A heuristic called Double Pivot (DP) is introduced which is an extension of the Single Pivot (SP) heuristic. It is shown that DP outperforms the Neighbor-Joining (NJ) heuristic, a method with wide acceptance within the biology community. The study includes many of the optimization criteria and indicates that form most of the criteria DP can find tree topologies with better values than the trees produced by NJ. The importance of using linear programming for choosing a good topology is shown and that local search does not improve the relative performance of NJ compared to DP.
Original language | English (US) |
---|---|
Pages | 98 |
Number of pages | 1 |
State | Published - 1997 |
Event | Proceedings of the 1997 1st Annual International Conference on Computational Molecular Biology, RECOMB - Santa Fe, NM, USA Duration: Jan 20 1997 → Jan 23 1997 |
Conference
Conference | Proceedings of the 1997 1st Annual International Conference on Computational Molecular Biology, RECOMB |
---|---|
City | Santa Fe, NM, USA |
Period | 1/20/97 → 1/23/97 |
ASJC Scopus subject areas
- General Engineering