【摘要】 本文提出了一种联想记忆网络权学习算法,文中将联想模型分解成一系列的非线性方程,针对这些非线性方程设计了一种能快速收敛的迭代方法,为了提高网络对缺损和噪声样本的联想记忆能力,通过提高网络阈值函数门限进行网络的严格训练,降低门限进行联想学习结果使得训练样本成为网络的稳定收敛点,并提高了网络的联想记忆能力。更多还原

【Abstract】 An effective weight, learning method for associative memory model is proposed in this paper. Associative memory model is divided into a series of nonlinear equations. Then a fast iterative method is devised for these nonlinear equations. In order to increase the capability for associative memory model to associate the correct information from the corrupted pattern, the threshold is suggested high at the training process and low at the associating process. The results of this learning method show that every trained pattern is a stable point and that the associating capability of the model is improved.更多还原