对随机取样的过程、特征及缺陷进行了分析，提出一种基于KD 树子样的聚类初始化方法。

This paper focuses on the process and property of random sampling, and proposes a novel random sampling method which is based on KD-Tree samples.

youdao

传统的模糊c -均值(FCM)聚类是一种基于梯度下降的优化算法，该方法对初始化较敏感，且易陷入局部极小。

The traditional fuzzy C-means (FCM) algorithm is an optimization algorithm based on gradient descending. it is sensitive to the initial condition and liable to be trapped in a local minimum.

youdao

为解决此问题，提出一种基于捕食-被捕食的粒子群优化模糊聚类算法且聚类中心采用密度函数初始化。

To solve the problem, a fuzzy clustering based on predator prey PSO algorithm is presented, which is using density function to initialize cluster centre.

youdao