摘要
聚类问题一直是模式识别和机器学习领域一个比较活跃而且极负挑战性的研究方向。谱聚类是近年来兴起的一类较流行的聚类方法。该文将非负约束引入到传统的谱聚类方法中,提出了一种基于非负约束的谱聚类方法。非负约束已在许多应用领域被证明是一种有用的性质。文中对比实验表明,基于非负约束的谱聚类方法在整体上明显优于传统的谱聚类方法。
Clustering is a challenging and active research topic in pattern recognition and machine learning.Spectral clustering is a new method for clustering.In this paper,the nonnegative constraint is introduced into the traditional spectral clustering.The NMF-based is proposed.The advantage of the nonnegative constrain has been conformed in many application fields.The results of the experiments in the paper evaluate the proposed method.
出处
《电脑知识与技术(过刊)》
2011年第6X期4165-4167,共3页
Computer Knowledge and Technology
基金
海南省教育厅高校科研项目(hjkj2010_50)
关键词
谱聚类
非负约束
聚类
非负矩阵分解
spectral clustering
nonnegative constraint
clustering
nonnegative matrix factorization