节点文献
基于结构动态可塑原理实现的神经网络感知器
Realization of the Perceptron with a Neural Network on the Principle of Dynamic Plasticity of a Structure
【摘要】 分析了人工神经网络感知器的基本性质,提出了一种新的硬件实现方法——结构动态可塑原理,即通过网络中连接路径的转换,间接实现权值系数的变化,达到学习的目的.并用一个逻辑功能感知器给出了具体的实现方法,证明该结论可推广到一般情况.
【Abstract】 The self-learning and self-tuning functions of the artificial neural network perceptron are realized by means of the dynamic change of weights. It is very easy to deal with the methematical aspect and simulation of weights, but it is difficult to realize the change in practice. This paper presents a realizing method on the principle of dynamic plasticity of a structure, that is, through the dynamic change of routes in the network to indirectly realize the change of weighting value. An example of a logic perceptron is given to show the method details.
【关键词】 神经网络感知器;
感知器实现方法;
结构动态可塑;
自学习;
自组织;
【Key words】 perceptron; realization of perceptron; dynamic plasticity of structure; self-learning; self-tuning;
【Key words】 perceptron; realization of perceptron; dynamic plasticity of structure; self-learning; self-tuning;
【基金】 国家自然科学基金资助项目
- 【文献出处】 华中理工大学学报 , 编辑部邮箱 ,1993年03期
- 【被引频次】1
- 【下载频次】44