摘要：文献[1-9]提出了血液循环在大脑处理信息的过程中具有时序控制作用，并用量化模型结合结构风险最小化相关理论说明时序控制作用的意义。文献[10-24]汇总介绍量化模型中的一些细节。本文介绍我们开发的一个算法，这一算法实现将一个有向网络分解为一系列前向网络集合。分解出来的前向网络集合可用于分析各种情况对任一细胞活动情况的影响，也可用于搭建精细的神经网络模型，进而用于辅助医学等方面的研究。算法的网络分解能力能符合文献[1-28]所介绍的大脑处理信息量化方案的要求

关键词 过程存储和重组模型；时序控制；脑电波；微循环；结构风险；中枢神经系统；信息处理；微环路；时间认知；智力起源；大脑量化模型；前向网络树

中图分类号：Q426        文献标识码：A           文章编号：            

An Arithmetic For Neural Network Analysis：From Directed Graph To FFN Trees

XIEQIN^1,*

1 Bureau of Science, Technology and Information of Guangzhou Municipality ; IT&T Department, GAGOC  Guangzhou，510000

Abstract: Literatures [1-9] suggest that blood circulation plays the role of basic timer when brain processing information; and suggest a quantitative model of brain information processing. Literatures [10-24] introduced details of the quantitative model.This article introduce an arithmetic that we design for network analyse.This arithmetic is able to change a directed network with feedback loops into a set of feedforward networks，according to the quantiative solution suggested by literatures [1-9]

Keywords: model of process storing and recalling; timing control; EEG; microcirculation; structure risk minimization; CNS; information processing; micro circuit; time cognition; origin of intelligence; quantitative model of brain information processing; FeedForward Network Tree; FFN Tree

一种可用于脑神经网络分析的有向图分解算法 第六稿.doc