摘要
基于牛顿前向插值公式提出一种对任意阶多维函数可实现无差逼近的新型CMAC联想记忆系统,详细讨论了该系统的原理、插值算法及训练规则,仿真实例表明了系统的可行性与有效性.该系统在信号处理、模式识别及高精度实时智能控制等领域具有广泛的应用价值.
A novel highorder associative memory system based on the Newton's forward Interpolation (NFIAMS) is preposed, which can carry out errorfree approximations to multivariable polynomial functions of arbitrary order. The theory, interpolation algorithm and traning rules of the NFIAMS are discussed in detail. The results of simulating examples indicate that it has more advantadges than CMACtype AMS, such as highprecision of learning, much smaller memory requirement without the datacollision problem and much less computational effort for training and faster convergence rates than that attainable with multilayer BP neural networks.
出处
《天津师范大学学报(自然科学版)》
CAS
2003年第2期55-57,共3页
Journal of Tianjin Normal University:Natural Science Edition
基金
天津市自然科学基金资助项目(013602811)