| 基于遗传算法的HMM参数估计 |
| |
| 引用本文： | 徐丽,康瑞华. 基于遗传算法的HMM参数估计[J]. 湖北工业大学学报, 2006, 21(4): 68-71 |
| |
| 作者姓名： | 徐丽 康瑞华 |
| |
| 作者单位： | 湖北工业大学计算机学院,湖北,武汉,430068 |
| |
| 摘 要： | 隐马尔可夫模型(Hidden Markov model)用于多序列比对研究是生物信息学研究的新领域,其可以通过训练识别同一特征的蛋白质序列.然而,目前的HMM参数估计算法Viterbi算法和Baum-Welch算法,都只能找到局部最优比对,无法找到全局最优比对.针对此算法全局最优问题提出了基于遗传算法的HMM参数估计,与已有的训练算法相比,遗传算法在搜索全局最优时具有突出的优势.
|
| 关 键 词： | 隐马尔可夫模型 遗传算法 多序列比对 |
| 文章编号： | 1003-4684（2006）08-0068-04 |
| 收稿时间： | 2006-05-23 |
| 修稿时间： | 2006-05-23 |
Traning of HMM With Genetic Algorithms |
| |
| XU Li,KANG Rui-hua. Traning of HMM With Genetic Algorithms[J]. Journal of Hubei University of Technology, 2006, 21(4): 68-71 |
| |
| Authors: | XU Li KANG Rui-hua |
| |
| Affiliation: | School of Computer, Hubei Univ. of Technology, Wuhan 430068, China |
| |
| Abstract: | Hidden Markov model used in multiple alignment is a new field of bioinformatics.It can be used to distinguish protein sequence with the same characterstics.But the problem is that the two parameter estimation algorithms for HMMs,the Viterbi algorithm and Baum Welch algorithm,are only guaranteed to find a local optimum.So a novel training method based on genetic algorithm is designed in this paper.Experimental results shows that this algorithm is more superior to those traditional methods in finding a global optimum. |
| |
| Keywords: | hidden markov model genetic algorithms multiple alignment |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
