基于多基因族编码的股票网络社团划分研究

发布时间：2018-07-04 11:35

本文选题：股票复杂网络 + 社团划分　；参考：《计算机工程与科学》2013年07期

【摘要】：针对传统股票网络社团划分算法发现精度低、时间复杂度高、容易陷入局部最优解的缺点,提出一种基于多基因族(MGF)编码的基因表达式编程(GEP)股票网络社团划分算法,来研究股票市场复杂网络社团化现象。该算法利用多基因族编码的特性,将代表股票节点的ID号和表示社团的类型分别编码在两个不同的多基因族中,再通过一个映射函数将两者的相互作用关系隐式编码在染色体中;同时,将精英迁移策略应用到基因选择、交叉、倒置、限制交换等各个遗传阶段,以避免早熟现象,加快遗传收敛到全局最优解的速度。实验分析表明,该算法能够准确和高效地实现股票复杂网络社团的划分,其划分结果对投资者进行决策具有重要的指导意义。
[Abstract]:Aiming at the shortcomings of traditional community partition algorithm of stock network, such as low precision, high time complexity and easy to fall into local optimal solution, a genetic expression programming (GEP) algorithm based on multiple gene family (MGF) coding is proposed. To study the stock market complex network corporatization phenomenon. In this algorithm, the ID number representing the stock node and the type representing the community are encoded in two different polygenic families, respectively. At the same time, the strategy of elite migration is applied to the genetic stages of gene selection, crossover, inversion, restriction exchange and so on, in order to avoid the phenomenon of precocity. Accelerate the speed of genetic convergence to the global optimal solution. The experimental results show that the algorithm can accurately and efficiently partition the complex network communities, and the division results are of great significance for investors to make decisions.
【作者单位】：江西理工大学信息工程学院;华南农业大学信息学院;
【基金】：国家自然科学基金资助项目(70971043) 江西省教育厅科学技术研究项目(GJJ12348) 广东省科技攻关项目(2012A020602037)
【分类号】：TP393.09

【相似文献】