基于联想记忆神经网络模型的BP算法  

An Improved BP Algorithm Based on Associative Memory Neural Network

作　　者：吴迪[1] 赵鹤鸣[1] 陶智[2] 

机构地区：[1]苏州大学电子信息学院,江苏苏州215006 [2]苏州大学物理学院,江苏苏州215006

出　　处：《微电子学与计算机》2005年第5期151-154,158,共5页Microelectronics & Computer

基　　金：国家自然科学基金资助(60172016);苏州大学青年基金资助(Q3108404)

摘　　要：提出一种改进神经网络BP算法的识别训练系统。实验结果表明,基于联想记忆神经网络模型的BP算法,大大降低样本可变系统的网络训练次数。该系统有效降低样本可变系统的网络训练时间,为BP算法应用于对实时性要求高的场合提供算法基础。In this paper, a Neural Network training system is proposed for improving BP algorithm. The BP algorithm training system is based on the improved BP algorithm and the information memory character of the Neural Network .The simulations show the BP algorithm based on Associative Memory Neural Network can greatly reduce the training time of the alteration swatch system .The system has efficiently decreased the training time of the alteration swatch system, and will provide algorithm elements for the Real-time systems.

关 键 词：BP算法 神经网络 联想记忆 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]