用线性神经网络提取主分量  

PRINCIPAL COMPONENT EXTRACTION BY LINEAR NEURAL NETWORK

作　　者：徐彤[1] 徐近霈[1] 

机构地区：[1]哈尔滨工业大学,150001

出　　处：《模式识别与人工智能》1994年第4期343-347,共5页Pattern Recognition and Artificial Intelligence

基　　金：国防科技预研基金

摘　　要：本文对用线性神经网络提取主分量的现有各种算法和作者提出的算法进行了分析和实验比较。模拟实验结果表明,所提出的算法以收敛的单调性、收敛速度较快、收敛精度较高而优于其它算法。Several existing algorithms and one improved algorithm used for principal component extraction by linear neural network are analyzed and compared experimentally in this paper. Simulation results are given to show the accuracy and speed of these algorithms, and the conclusion is in favour of the proposed algorithm for its monotonous convergence property, fast convergence speed and higher computational accuracy.

关 键 词：主分量提取 神经网络 图像处理 模式识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]