神经元网络在SAS中的实现(Ⅰ)

大多数常见的人工神经元网络—多层感知器（ＭＬＰ），无非是非线性回归以及判别模型。其训练方法通常是适合大型并行机硬件实现的梯度下降算法。对于一般的计算环境，ＳＡＳ系统中若干优化算法远比神经元网络的并行算法更有效。本文解释神经元网络是什么，描述神经元网络与统计模型的关系，介绍神经元网络在ＳＡＳ中的实现。

The Implementation of Neural Network Application in SAS

AbstractMost commonly used artificial neural networks are nothing more than nonlinear regression and discriminant models that can be implemented with standard statistical software. This paper shows the relations between neural networks and statistical models and the implementation of Neural Network Application in SAS.