基于SNP系统的改进粒子群聚类算法

Improved Particle Cluster Algorithm Based on SNP System

脉冲神经膜系统是基于神经生物学的高性能计算模型.在标准粒子群聚类算法中引入脉冲神经膜系统,将初始聚类中心的各种组合作为粒子分配到若干个神经元,在神经元中进行粒子群的迭代与进化.利用脉冲神经膜系统的高并行性,在更短的时间内得到更优化的初始聚类中心,为K-means算法的局部寻优提供更好的聚类初值.实验结果表明,改进后的算法可以进一步提升聚类的准确率,取得更好的聚类效果.

Spiking neural P(SNP) systems are high-performance computational models based on neurobiology.SNP systems are introduced into the standard particle swarm clustering algorithm.Various combinations of initial clustering centers are assigned to several neurons as particles,and the particle swarm iteration and evolution are carried out in the neurons.The high parallelism of SNP systems is utilized.A more optimized initial clustering center is obtained in a shorter time,which provides a better initial clustering initial value for the local optimization of K-means algorithm in the next step.Experimental results show that the improved algorithm can further improve the accuracy of clustering and achieve better clustering results.