一种新的高阶联想存贮模型

A Nevel High-order Associative Memory Model

由于Kohonen模型对噪声极端敏感，Murakami提出一个利用带噪输入优化Kohonen模型的最小平方联想存贮模型（LSAM），大大降低了原有模型的噪声敏感性。但是与Kohonen模型一样，LSAM模型的联想存贮能力随着样本数的增加而大大下降。本文提出一种高阶联想存贮模型，对LSAM模型作了改进，使原有模型的联想存贮能力得到极大地提高。计算机模拟结果证实了这一点。

Due to noise sensitivity of Kohonen's model, Murakami proposed a least squares associative memory model (LSAM) by using noise-corrupted imputs to optimize kohonen's model. so that greatly reduces the noise sensitivity of the original model. However. like Kohonen's model, the associative performence of LSAM decrease with the increase of the number of samples. In this paper, a novel high-order associative memory model is proposed, it improves the LSAM and greatly raises the associative performence of the former. Finally, computer simulation verity its advantege.