摘要
提出一种基于信息熵的蚁群聚类算法,将信息熵引入到LF算法中,数据对象的归属由信息熵来决定,减少了参数,测试并验证了算法的有效性;同时,针对信息熵的蚁群算法早期数据分散、收敛过慢、容易陷入局部最优等缺点,提出了一种蚁群聚类组合方法。改进思路是引入K-means作为熵蚁群算法的预处理过程,通过K-means快速、粗略地确定聚类中心,利用K-means方法的结果作为初值,再进行改进的熵蚁群算法聚类,有效地解决了蚁群算法早期收敛过慢等问题。
Proposed a new ant colony clustering based on information entropy,introduced the entropy into the LF algorithm,which determined the state of the data,and reduced the parameters to test and verify the effectiveness of the algorithm.At the same time,for the information entropy of ant colony algorithm＇s early data were too scattered so convergence was slow.Vulnerable to the shortcomings of local optimum,presented a combination method to improve the ant colony clustering.The paper introduced K-means to the pre-computation process of ant colony algorithm.Through K-means,it determined cluster center fast and sketchily,and got the starting value using the K-means result,then clustered by the improved algorithm.It effectively solve the slow convergence of ant colony algorithm for the early issues.
出处
《计算机应用研究》
CSCD
北大核心
2011年第4期1269-1271,共3页
Application Research of Computers
基金
辽宁省自然科学基金资助项目(20082002)
关键词
聚类
蚁群聚类
信息熵
K-均值
clustering
ant colony clustering
information entropy
K-means
