基于自组织竞争神经网络的红籽瓜自交系聚类分析

The cluster analysis of red-seed watermelon inbred line based on self-organizing competitive neural networks

自组织竞争神经网络是一种具有强大的自学习功能、良好的自组织性和自适应性的聚类方法,能够迅速得到聚类结果。本文基于自组织竞争神经网络的聚类功能,依据6个红色籽用西瓜数量性状特征指标,将47份自交系分为4类。聚类结果表明,不同自交系类群间的遗传变异较大,在配置杂交组合时,应尽量在不同类群间选配亲本,为红籽瓜育种提供理论依据。

Self-organizing competitive neural networks is a kind of stong self-lerning, self-organizing and self-suitable cluster method, which can quickly gain the cluster result. Based on the cluster function of self-organizing competitive neural networks and six characters of red-seed watermelon, 47 inbrde lines are divided into 4 species. The cluster results indicate that the different clusters have more inheritance variation. We ought to choose parents from different clusters when configuring hybrid combination, which provides theory basis for red-seed watermelon breeding.