基于混合生物地理学优化的聚类算法

聚类分析是数据挖掘的重要任务之一,而具有易早熟与收敛速度慢等缺陷的传统生物地理优化算法(Biogeography-Based Optimization, BBO)很难满足具有NP(Non-deterministic Polynomial)性质的复杂聚类问题需求,于是提出了一种基于混合生物地理学优化的聚类算法(Mixed Biogeography-Based Optimization, MBBO),该算法构造了一个基于梯度下降局部最优贪心搜索的新迁移算子,以聚类目标函数值作为个体适应度进行数据集内隐簇结构寻优。通过在4个标准数据集(Iris、Wine、Glass与Diabetes)的实验,结果表明MBBO算法相对于传统的优化算法具有更好的优化能力和收敛度,能发现更高质量的簇结构模式。

Mixed Biogeography-Based Optimization Algorithm for Data Clustering

Cluster analysis is an important task of data mining, however, traditional biogeography-based optimization algorithm with limitations such as prematurity and poor convergence can not satisfy the demands of solving the NP (Non-deterministic Polynomial) clustering problem. A novel algorithm Mixed Biogeography-Based Optimization (MBBO) is proposed. The algorithm integrates a new migration operator, which is constructed on gradient descent local search, and uses clustering validity index as the individual fitness to optimize implicit cluster structures in datasets. Experimental results on the four benchmark datasets (Iris, Wine, Glass and Diabetes) show that MBBO algorithm outperforms the traditional optimization algorithms such as PSO, BBO, and K-means in terms of clustering validity and convergence, and can acquire the higher quality cluster structures of the datasets.