神经记忆的电阻网络综合
RESISTIVE NETWORK SYNTHESIS FOR NEURAL MEMORY
-
摘要: 本文提出了从神经系统功能进行网络综合的观点。对神经网络的记忆功能进行了研究,建立了数学模型,并用分段线性电阻网络进行综合。所得到的非线性电阻网络除具有并行性、实时性、分布性和自适应等一系列人工神经网络的特点外,网络中各参数由网络所应记忆模式和特征向量的解析式给出,这比传统的人工神经网络具有较强的直观性,便于实际应用。
-
关键词:
- 神经网络; 神经记忆; 电阻网络综合
Abstract: It is suggested to study the neural network in the view of neural functions. The memory function of network is studied and its mathematical model is given. The model is synthesized by a peicewise-linear resistive network. The network is of many properties of artificial network such as parallelism, real-time processing, distributivity and adaptability etc., in addition, the parameters of network are expressed analytically by the patterns and features which are memorized in the network. -
R.P. Lippmann, IEEE Trans. on ASSP, ASSP-4(1987)4,4-22.[2]夏征农等,辞海(1979年版缩印本),上海辞书出版社,1985年,p.380.[3]Jack Y. Jau, Y.Fainman, Sing H. Lee, Applied Optics, 28(1989)2, 302-305.[4] B.Kosko, IEEE Trans. on SMC, SMC-18(1988)1,49-60.[4]蔡少棠, 非线性网络理论引论(中册),人民教育出版社,1980年.
计量
- 文章访问数: 1899
- HTML全文浏览量: 128
- PDF下载量: 412
- 被引次数: 0