基于多输入泛函网络的构造和学习策略  

Strategy of Structuring and Learning Based on Multi-Input and Single Output Functional Network

作　　者：崔明义[1] 张新祥[1] 苏白云[1] 张瑞[1] 

机构地区：[1]河南财经学院计算机科学系,郑州450002

出　　处：《计算机科学》2006年第10期169-171,192,共4页Computer Science

基　　金：河南省自然科学基金(0411013800;0411014500);河南省高校杰出科研人才创新工程项目(2004KYCX014)资助

摘　　要：泛函网络是类似于人工神经网络的新型网络模型,是泛函方程的网络表达形式。本文针对复杂泛函网络构造和学习中存在的问题,提出了多输入泛函网络模型MIOFN。通过对该模型的分析,提出了简化和学习的方法,并进行了仿真实验。结果表明,本文提出的MIOFN运行是可靠的,在工程应用中是有效可行的。Functional network is new network model. It is similar to artificial neural network and is network expression of functional equation. In this paper, a multi-input and single output FN were presented for solving problems in structuring and learning of complex FN. Through the model was analyzed, a new simplifying and learning method was put forward by it. The Simulating experiment was done by it. The result indicates that running of the model is reliable. It is applied to engineering availably and feasibly.

关 键 词：泛函网络 MIOFN 泛函方程 拓扑结构 学习 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]