量子球壳聚类  

Spherical shell clustering based on quantum measurement

作　　者：杨勋[1] 谢维信[1] 黄建军[2] 

机构地区：[1]西安电子科技大学电子工程学院,陕西西安710071 [2]深圳大学ATR国防科技重点实验室,广东深圳518060

出　　处：《西安电子科技大学学报》2008年第1期43-48,共6页Journal of Xidian University

基　　金：国家自然科学基金资助(60572102);深圳市科技计划项目资助(200506)

摘　　要：提出一种基于量子测量的球壳聚类方法.利用量子测量机理来部分模拟人脑有意识的认知机制,其基本思想是将聚类样本集分别视为可测量的环境量子系综和刺激量子系综,每个样本视为宏观系综中的一个微观量子叠加态.刺激量子系综依概率在具体微观态上进行塌缩产生刺激引起无意识的注意,并产生球壳面算符.球壳面算符通过对环境量子系综进行有意识的测量并得到平均测量值来验证球壳面的有效性,同时可以得到壳面上的样本集.相对模糊球壳聚类方法,量子球壳聚类方法能有效初始化和自动确定聚类数目,其所体现的融合信息加工过程中无意识数据驱动和有意识概念驱动的量子认知思想有望在其它模式提取方面作进一步的积极扩展应用.This paper proposes a method based on quantum measurement for spherical shell clustering. This method simulates the conscious cognition of the human brain using quantum measurement. It regards a sample set as an environment quantum ensemble which is measurable and the stimulus quantum system. In this thought, each sample is treated as a microcosmic state of quantum superposition. The stimulus quantum ensemble attracts the conscious attention and makes the cognition mechanism produce corresponding spherical shell surface operators by collapsing on microcosmic states in probability. A shell surface operator is applied to measure the environment quantum ensemble and the average measure value can be obtained which validates the operator. At the same time, a sample set on the shell surface will come forth. Quantum shell clustering has the ability of perfect initialization and auto-detection of the class number of clustering compared with fuzzy shell clustering. The quantum cognition thought,which integrates unconscious data-driving and conscious concept-driving in information-processing, can be further studied and applied in the other pattern analysis.

关 键 词：球壳聚类 球壳面算符 量子测量 量子系综 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]