共查询到18条相似文献,搜索用时 625 毫秒
1.
2.
提出了设计联想记忆神经网络的模式分量最优稳定性准则,即使得每个模式分量对于模式畸变都具有最大的联想容错性。 相似文献
3.
具有轴突信号传输延迟的双向联想记忆神经网络 总被引:8,自引:1,他引:7
双向联想记忆模型是两层异联想网络,本文重点讨论了具有轴突信号传输延迟双向联想记忆神经网络平衡点存在的条件以及全局稳定性,并用具体例子验证了结论的正确性,本文所讨论的方法具有普遍性,所得到的结果对于网络的设计和应用具有重要的指导意义。 相似文献
4.
5.
最大—乘积型模糊联想记忆网络的最大最小编码学习算法 总被引:4,自引:0,他引:4
提出了最大-乘积型模糊联想记忆网络的最大最小编码学习算法,新算法可以记忆任意多个自联想模式,对于异联想模式,给 种以最大最小编码算法为基础,近似求解网络连接权阵的梯度下降学习算法,这种方法可用于解最大乘积型模糊关系方程,计算机模拟实验证有效性。 相似文献
6.
7.
8.
综合联想记忆神经网络的外积取等准则 总被引:8,自引:1,他引:7
本文提出了一个新的联想记忆设计准则,即外积取等准则,它具有外积和准则的所有优点。由外积取等准则设计出的联想记忆网络能够存储任意给定的训练模式,即对于训练模式的数目和它们之间相关性的强弱没有限制。外积取等准则可用来定量地评价记忆模式向量各分量对于记忆模式分类或识别的重要性。由外积取等准则设计出的网络的连接权值只取1、0或-1,因而网络易于光学实现。计算机实验结果充分说明了外积取等准则的有效性。 相似文献
9.
Hopfield连续联想记忆的吸引域和指数收敛速度的估计及其应用 总被引:10,自引:0,他引:10
本文得到了Hopfield连续联想记忆各记忆模式的吸引域及其中每一点趋向记忆模式的指数收敛速度的估计结果,它们可用来评价Hopfield连续反馈联想记忆的容错能力,且可用于综合更为实用的连续联想记忆神经网络。 相似文献
10.
11.
A method based on fuzzy associative memories for calculating the bandwidth of electrically thin and thick rectangular microstrip patch antennas is presented. The design parameters of the fuzzy associative memories are optimally determined by using the classical, modified and improved tabu search algorithms. The theoretical bandwidth results obtained by using the fuzzy associative memories are in very good agreement with the experimental results available in the literature. 相似文献
12.
双向联想记忆神经网络的一种编码策略 总被引:4,自引:0,他引:4
本文提出一种双向联想记忆神经网络的按‘位’加权编码策略,并给出了求取权值递推算法,它将Kosko双向联想记忆网络按海明距离进行模式匹配的原则,修正了按加权海明距离进行了模式匹配,从而可以使得对不满足加续性的所谓“病态结构”的一类样本模式集,同样具有良好的联想能力,对二值图象模式存贮,联想的计算机模拟实验表明,此方法具有优良的性能和实用价值。 相似文献
13.
Xu Shundou 《中国邮电高校学报(英文版)》1997,(1)
OntheDesignofNonlinearNeuralNetworksforAsociativeMemoriesXuShundou(DepartmentofBasicCourses,ShijiazhuangPostalColege,Shijiazh... 相似文献
14.
Algorithm and Architecture of Fully-Parallel Associative Memories Based on Sparse Clustered Networks
Hooman Jarollahi Naoya Onizawa Vincent Gripon Warren J. Gross 《Journal of Signal Processing Systems》2014,76(3):235-247
Associative memories retrieve stored information given partial or erroneous input patterns. A new family of associative memories based on Sparse Clustered Networks (SCNs) has been recently introduced that can store many more messages than classical Hopfield-Neural Networks (HNNs). In this paper, we propose fully-parallel hardware architectures of such memories for partial or erroneous inputs. The proposed architectures eliminate winner-take-all modules and thus reduce the hardware complexity by consuming 65 % fewer FPGA lookup tables and increase the operating frequency by approximately 1.9 times compared to that of previous work. Furthermore, the scaling behaviour of the implemented architectures for various design choices are investigated. We explore the effect of varying design variables such as the number of clusters, network nodes, and erased symbols on the error performance and the hardware resources. 相似文献
15.
Zhang Yongjun 《电子科学学刊(英文版)》1996,13(1):11-16
This paper provides a new architecture of neural network, called loop architecture neural network(LANN), and its learning rules. One of its features distinguished from other network, such as Hopfield and bidirectional associative memories, is that it can perform the associative memory among multiple categories. Analysis and simulated results have proved that it is an effective network with excellent convergence. 相似文献
16.
Dembo A. 《IEEE transactions on information theory / Professional Technical Group on Information Theory》1989,35(4):709-720
Some important features of various known constructions of associative memories based on linear threshold functions are analyzed. Two important features are dealt with: (a) the ability to select an arbitrary set of desired memory vectors and design a network for this set; (b) the sizes and shapes of the domains of attraction of the desired memory vectors and their relation to various design parameters. The static capacity for randomly chosen desired memories is also analyzed. Two extremal examples of sets of desired memories are then analyzed in detail. For spectral schemes with randomly chosen O[N /ln N ] memories, it is shown that almost all of the Hamming sphere around each memory is directly attracted 相似文献
17.
模糊联想记忆的一种有效学习算法 总被引:13,自引:0,他引:13
本文提出最大-最小合成模糊联想记忆的一种新的学习算法,在一定条件下,这种学习算法能将多个模糊模式对可靠地编码到模糊联想记忆的连接权矩阵中,且已存储的模式对可被完整的回想出来。 相似文献