首页 | 本学科首页   官方微博 | 高级检索  
     检索      

离散型 Hopfield 联想存储器k阶吸引子的鲁棒性分析
引用本文:胡东成,杨良土,罗予频.离散型 Hopfield 联想存储器k阶吸引子的鲁棒性分析[J].清华大学学报(自然科学版),1998(9).
作者姓名:胡东成  杨良土  罗予频
作者单位:清华大学自动化系
基金项目:国家自然科学基金,国家教委高等学校博士点基金
摘    要:为对离散型Hopfield联想存储器(以下简称为DHAM)平衡点吸引域的鲁棒性进行理论探索,文章基于输入及权值实现误差或扰动同时作用的随机模型,采用统计学方法,推出了DHAM的k阶吸引子(k≥0)鲁棒性的通用计算方法。并针对正交样本经Hebb规则构成的DHAM,进一步得到了该类网络k阶吸引子鲁棒性的具体计算公式。仿真试验表明所得算法是正确的。讨论了这类DHAM结构及参数等对其k阶吸引子鲁棒性的影响。

关 键 词:联想存储器  实现误差  吸引子  鲁棒性

Robustness analysis of k order attractors in discrete Hopfield associative memory
HU Dongcheng,YANG Liangtu,LUO Yupin.Robustness analysis of k order attractors in discrete Hopfield associative memory[J].Journal of Tsinghua University(Science and Technology),1998(9).
Authors:HU Dongcheng  YANG Liangtu  LUO Yupin
Institution:HU Dongcheng,YANG Liangtu,LUO Yupin Department of Automation,Tsinghua University,Beijing 100084,China
Abstract:The robustness of attractive equilibria which have been stored in Discrete Hopfield Associative Memory (DHAM) is studied. Based on a stochastic model for input and weight implementation errors which may exist simultaneously, a general algorithm is developed to compute the robustness of k order attractors in a given DHAM by using statistical approach. The formula is given to calculate the robustness of k order attractors in DHAM trained by Hebb law upon mutually orthogonal patterns. The simulation results indicate the correctness of the proposed algorithm. The influence of structure and parameters of this kind of DHAM on the robustness of its k order attractors is discussed.
Keywords:
本文献已被 CNKI 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号