摘要
文章提出一种过程神经元模型 ,其输入为与时间有关的函数或过程 ,它是传统人工神经元模型在时间域上的扩展。基于这种过程神经元模型 ,给出了一种仅含一个隐层的前馈型过程神经网络模型 ,即基展开过程神经元网络模型。证明了相应的连续性定理 ,逼近定理 ,计算能力定理等。
In this paper, a novel artificial neuron model procedure neuron model is proposed, in which the inputs are functions or procedures associated with `time'. Based on these neurons, a model named procedure neural network, which is also a feedforward network with only one hidden layer, is constructed. The authors call this neural network as Procedure Neural Network (PNN) expanded on certain base functions. The related continuity, function approximation ability and computational capability theorems are proved. [
出处
《中国工程科学》
2000年第12期40-44,共5页
Strategic Study of CAE
关键词
过程神经元网络
函数逼近能力
计算能力
连续性
网络模型
procedure neural networks
function approximation ability
computational capability
continuity
