A Logical Analysis for Stability of Neural Networks

Guo Baolong

您当前的位置：

首页 >

文章列表页 >

A Logical Analysis for Stability of Neural Networks

更新时间：2025-12-08

- A Logical Analysis for Stability of Neural Networks
- Acta Electronica Sinica Issue 11, (1996)
- 作者机构：
  
  1. 西安电子科技大学检测与仪器系
  2. 西北工业大学自动控制系
  3. 西安电子科技大学检测与仪器系西北工业大学自动控制系
- 作者简介：
- 基金信息：
- DOI：
  CLC： TP18
- Published：1996
- 稿件说明：
移动端阅览
[1]郭宝龙,郭雷.神经网络稳定性的逻辑分析[J].电子学报,1996(11):1-5.

Guo Baolong. A Logical Analysis for Stability of Neural Networks[J]. Acta Electronica Sinica, 1996, (11).
[1]郭宝龙,郭雷.神经网络稳定性的逻辑分析[J].电子学报,1996(11):1-5. DOI：

Guo Baolong. A Logical Analysis for Stability of Neural Networks[J]. Acta Electronica Sinica, 1996, (11). DOI：

摘要

神经网络稳定性分析的传统方法是采用权矩阵的定量分析。本文提出一种基于推理和矛盾分析的逻辑分析法，并阐明导致网络不稳定的局部结构因素，这有利于神经网络的分析和设计。本文首先给出了二值神经网络（ＢＮＮ）的逻辑分析理论，然后对ＢＮＮ的稳定性给予定性分析，最后分别讨论了全约束规则和半约束规则形成矛盾环的情况。分析表明：神经网络不稳定的必要条件是网络中存在矛盾环。因此，有可能通过分析局部结构来研究网络的整体行为。

Abstract

Traditional approaches for the analysis of stability of neural networks are quanti tative such as analysis of weight matrixes.This paper presents a logical method based on inference and contradiction analysis and reveals what partial/local structures are unstable factors

which is useful in the design and analysis of neural networks.In the paper

a logical theory of binary neural networks（BNNs） is first presented.Then

the qualitative analysis of the stability of BNNs is given.Finally

the formation of contradiction loops with full-constraint and semi-constraint rules is discussed.The analysis indicates that the existence of contradictory loops inside BNNs is a neces sary condition for BNNs to be unstable

so that it is possible to study the global properties of net works by only analyzing their partial/local structures.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Neural Network for Solving Systems of Nonlinear Equations

A New Analytical Method for Hopfield Type Associative Memory Neural Networks

Convergence Analysis on Neural Network for Eigenproblem

A CN-RK Discontinuous Galerkin Time-Domain Method Without the Stability Restriction Due to Material Loss

Image Classification Network of Background Perception Mechanism

Related Author

ZHAO Hua-min

CHEN Kai-zhou

Mlao Zhenjiang

Yuan Baozong

叶仲泉

曾长修

XIANG Ru

MA Xi-kui

Related Institution

School of Science,Xidian University

Institute of Information Science,Northern Jiaotong University

北方交通大学信息所北京 100044

重庆大学系统工程及应用数学系

重庆大学自动化系

⁰