包含奇异类样本的感知机学习规则的单位圆算法

Unit Circle Algorithm of Perceptrons Study Rule with Oddity Sample

分析了传统的感知机学习方法的不足,通过对包含奇异类样本的分析,提出了基于矢量的零偏置的感知机学习规则:单位圆算法,并证明了该算法的收敛性。最后给出了实例样本的模式识别的实现及模式识别的结果。

This paper analyses the drawbacks of traditional perceptrons study rule in the first. Then it suggests a perceptrons study rule based on vector and zero bias through the analyses with oddity sample-unit circle algorithm. It also proves the algorithm's astringency. Finally, it gives experimental data and diagram between traditional study rule and unit circle algorithm for pattern recognition.