摘要
建立了基于ALM和权值的LRR聚类改进模型,对高维数据进行分析,将其分为两个独立的子空间,并与传统k-means聚类模型进行对比,采用评价指标模型对聚类结果进行评价分析.提出的LRR聚类改进模型在正则项引入了权重系数w,可以更好地将扰动分开,求解结果及评价指标均有效地验证了其稳定性、精确度等性能均有所提升.建立了SMMC改进模型,对机器工件外部边缘轮廓进行分类.从求解结果可看出该模型非常适合用于处理混合多流形聚类问题,对于比较复杂的曲线有着很好的分类性能.按照数据预处理、数据建模分析、模型结果评价步骤,通过使用谱聚类分析和多流形学习方法,对所给出的高维数据进行分析和处理,并通过评价模型得出相应的评价指标,对数据的多流形结构进行了深入的研究和探讨.
This paper establishes improved LRR clustering model based ALM and weights to analyze the high-dimensional data in order to divide them into two separate sub-space. Compared with the traditional k-means clustering model, using evaluation model for the analysis of clustering results. The improved LRR clustering model introduces weighting coefficient in the regularization term, which can better separate the disturbance. The results and evaluation can effectively verify that its stability, accuracy and other properties have been enhanced. This paper establishes improved SMMC model for the classification of outer edges in the machine workpiece contour. As can be seen from the results that model is well suited for the treatment of mixed multi-manifold clustering problem and also has a good classification performance for more complex curves. In this paper, according to data preprocessing, data modeling and analysis, model results evaluation step, by using spectral clustering analysis and multi-manifold learning method, analyze and process high dimensional data presented, then draw the appropriate evaluation through the evaluation model, finally carry out in-depth study and discussion for multi-manifold structure of data.
出处
《数学的实践与认识》
北大核心
2016年第14期200-207,共8页
Mathematics in Practice and Theory
关键词
谱聚类
稀疏子空间
多流形学习
评价指标
spectral clustering
sparse subspace
multi-manifold learning
evaluation indicators
