产生Tuned模板的基因扰动优化法研究与分析

A Research and Analysis of Gene Disturb Optimization MethodWhich Produce Tuned Template

  • 摘要: 提出一个基于基因优化产生Tuned模板的新方法——基因扰动优化法(GDO法)。首先在初始模板群体中,根据求得的适应度值较好的少数几个模板,确定为优秀模板;然后对优秀模板中的每个基因分别进行扰动,根据扰动前后适应度值增加(或减少),确定每个优秀模板的两个优秀基因;最后,对优秀基因的数值引入扰动,得到一组新模板,进入下一代的计算,如此不断地优化得到最佳Tuned模板。经理论分析和对真实图像分类的实验证明,GDO法是一个很有潜力的优化方法。

     

    Abstract: This paper presents a new method that is based on genetic optimization which produce tuned template, called genedisturb optimization method (referred GDO method). Firstly, identified the "excellent template" according to a few templates of better fitness value which is obtained in the initial templates group. Then disturbing excellent template for each gene respectively, according tothe value of fitness before and after the disturbance increases (or decreases) to determine two "excellent genes" about each excellent template. Finally, disturb the numerical value of the excellent genes to get a new set of templates, then go into the next generation of calculation, constantly optimize the best tuned template. The theory analysis and the real image classification experiment proved that GDO method is a very potential optimization method.

     

/

返回文章
返回