###
DOI:
Journal of Software:1997.8(8):622-629

二进制神经网络分类问题的几何学习算法
朱大铭,马绍汉
(山东大学计算机系,济南,250100; 中国科学院计算技术研究所,北京,100080)
A GEOMETRICAL LEARNING ALGORITHM OF BINARY NEURAL NETWORKS FOR CLASSIFICATION
ZHU Daming,MA Shaohan
()
Abstract
Chart / table
Reference
Similar Articles
Article :Browse 2883   Download 3438
    Revised:September 17, 1996
> 中文摘要: 分类问题在前向神经网络研究中占有重要位置.本文利用几何方法给出一个二进制神经网络K(≥2)分类问题的新学习算法.算法通过训练点的几何位置与类别分析,建立一个四层前向神经网络,实现网络输入向量分类.本文算法的优点在于:保证学习收敛且收敛速度快于BP算法及已有的其他一些前向网络学习算法;算法可以确定神经网络的结构且能实现精确的向量分类.另外,算法所建神经网络由线性阀值单元组成,神经元突触权值和阀值均为整数,特别适合于集成电路实现.
中文关键词: 神经网络  算法  收敛  训练  几何
Abstract:Binary to binary mapping for classification plays an important role in the researches on feed-forward-neural-network learning.In this paper,the geometrical method is employed to work out a new algorithm to train binary neural networks for classification.By analysis of every training vertex's geometrical location,the algorithm alwavs produces a neural network of four layers for a certain classification problem.The advantages of this algorithm are:it runs with guaranteed convergence and goes to converge much more quickly than BP and some other algorithms;it can determine the structure of the neural networks by learning SO that a precise classification is carried out.In addition,every neuron generated by the algorithm employs a hard-limit activation func-tion with integer synaptic weights,which makes the actual implementation by VLSI tech-nology more facilitated.
文章编号:     中图分类号:    文献标志码:
基金项目:本文研究得到国家自然科学基金、国家863高科技项目基金和山东省自然科学基金资助. 本文研究得到国家自然科学基金、国家863高科技项目基金和山东省自然科学基金资助.
Foundation items:
Reference text:

朱大铭,马绍汉.二进制神经网络分类问题的几何学习算法.软件学报,1997,8(8):622-629

ZHU Daming,MA Shaohan.A GEOMETRICAL LEARNING ALGORITHM OF BINARY NEURAL NETWORKS FOR CLASSIFICATION.Journal of Software,1997,8(8):622-629