基于密度-距离的t混合模型流式数据聚类

发布时间：2018-05-13 00:12

本文选题：t-混合模型 + 密度-距离中心算法　；参考：《仪器仪表学报》2017年09期

【摘要】：传统流式数据采用人工设门法分析,效率低下且依赖于专家。近几年,很多自动流式数据聚类算法纷纷被提出,然而针对数据量不多且分布稀疏的小样本类群始终没有很好的解决办法。提出了一种基于密度-距离的t-混合模型流式数据聚类优化方法,能够较好地解决小样本类群区分困难的问题。该方法通过密度-距离中心算法定位各类群的初始中心,作为t-混合算法的初值对样本数据进行处理,通过最大似然估计求出各类群对应的样本数目,从而实现样本聚类。实验表明,与经典模型算法相比,基于密度-距离的t-混合模型优化算法具有更好的稳定性和可靠性,对小样本类群以及混叠的类群具有较强的适应能力。
[Abstract]:The traditional flow data is analyzed by manual gate setting, which is inefficient and depends on experts. In recent years, many automatic flow data clustering algorithms have been proposed. However, there is no good solution to the small sample cluster which has a small amount of data and sparse distribution. In this paper, a density-distance based clustering optimization method for t- hybrid model flow data is proposed, which can solve the problem of small sample clustering. This method uses the density-distance center algorithm to locate the initial center of various groups and processes the sample data as the initial value of the t- hybrid algorithm. The maximum likelihood estimation is used to calculate the number of samples corresponding to the various groups, so that the sample clustering can be realized. The experimental results show that the density distance based thybrid model optimization algorithm has better stability and reliability than the classical model algorithm, and has a strong adaptability to small sample groups and aliasing groups.
【作者单位】：上海大学机电工程与自动化学院;上海市智能制造及机器人重点实验室;上海纳衍生物科技有限公司;
【基金】：上海市浦江人才计划(17PJ1432300)项目资助
【分类号】：TP311.13

【相似文献】