基于改进随机云滴法的云相似性计算

现有随机云滴法得出的相似度随云滴数目的增加而存在趋势性变化,以致无法给出较为稳定、合理的计算结果。为此提出了一种改进方法,对随机云滴法出现的趋势性变化现象进行了分析和讨论,给出了相关定理及证明过程。通过探讨问题产生的实质,忽略云滴数目,将相似度统一建立在云滴距离这一物理概念上,重新定义了相似度表达式。仿真实验的结果表明,改进方法能够获得相对稳定的相似度计算结果,有效解决了原算法相似度的趋势性变化问题,具有较好的适用性。

Cloud similarity computation based on an improved randomdrops method

The similarity of current random drops method is characterized by single trend with increasing drops, causing less stable and reasonable results. Thus an improved method is proposed to solve this problem. Single trend phenomenon in random drops method is analyzed and discussed, and relevant theorem and corresponding demonstration are provided. Meanwhile, the number of drops in similarity is ignored by investigating the above problem, and the similarity formula is defined again based on the identical physical conception of drop distance. Simulate results indicate that our improved method can solve the problem of single trend with relatively stable similarity, and is more suitable for application.