摘要
在讨论数字逻辑与神经元的关系后，提出一种利用前向三层神经网络实现任意布尔逻辑的设计方案。并引入卡诺图化简及最小项抑制的思想降低神经网络的实现复杂性。文中给出设计的原理及算法流程图。此方案结构简单可靠、学习速度快、易于硬件实现。
After discussing the connection between digital logic and a neuron,a strategy of implementing arbitrary Boolean logic using three layers feedforward neural network is presented.The ideas of Karnaugh map and minterm inhibition are introduced to reduce the complexity of Boolean neural network.The principle and implementation steps of the algorithm are described in detail .The structure of this kind of network is simple ,reliable and easy to be implemented with hardware.Moreover the learning speed of the algorithm is very fast.
出处
《电路与系统学报》
CSCD
1998年第3期51-58,共8页
Journal of Circuits and Systems
关键词
数字逻辑
神经网络
卡诺图
布尔函数
Digital logic
Neural network
Karnaugh map
Boolean function
