摘要
提出一种快速算法,该算法利用贪心算法构造卷数据降维矩阵,在保持点与点之间"核距离"不变的情况下,把待分解矩阵变换成一个低维矩阵。在没有偏差的情况下,将对原始大矩阵的分解变成对这个低维矩阵的分解,大幅降低了时间复杂度,减少了对内存的使用率的同时增加了算法的稳定性。
The authors present a fast algorithm that uses the greedy algorithm to make wrapped dimensionality reduction data matrix in order to keep ＂nuclear distance＂ between points unchanged. The original matrix needed to be decomposed has been transformed into a low-dimensional matrix in the case of no deviation. Therefore we significantly reduce the time complexity and memory usage while increasing the stability of the algorithm.
出处
《微型机与应用》
2013年第19期72-75,78,共5页
Microcomputer & Its Applications
基金
国家自然科学基金项目(61105085)
关键词
主成分分析
贪心算法
卷数据降维矩阵
时间复杂度
PCA（Principal Component Analysis）
greedy algorithm
wrapped dimensionality reduction data matrix
time complexity
