论文部分内容阅读
递归相关联想记忆（ＲＣＡＭ）的回忆规则不同于Ｈｏｐｆｉｅｌｄ网络之处在于前者在输入与记忆模式的相关值上作用一非线性函数．在文献［７］的基础上，文中对所涉及的非线性函数进行了进一步的研究，提出了利用截断较小相关值来提高记忆性能的方法，得到了一种新的具有ＲＣＡＭ结构的联想记忆器（ＴＲＣＡＭ）．理论分析表明该方法可大大地提高记忆器对任意输入的信噪比，仿真实验也显示此方法可显著增大记忆模型在保证一定纠错能力下的记忆容量．
Remembrance associative memory (RCAM) recall rules differ from the Hopfield network in that the former operates a nonlinear function on the correlation of input and memory modes. Based on the literature [7], the paper further studies the nonlinear functions involved and proposes a method to improve the memory performance by truncating the smaller correlation values. A new associative memory with RCAM structure is obtained Device (TRCAM). Theoretical analysis shows that this method can greatly improve the signal to noise ratio of the memory to any input, and the simulation experiment also shows that this method can significantly increase the memory capacity of the memory model with certain error correction capability.