论文部分内容阅读
1、神经网络的实现方法神经网络的实现方法有多种多样,对这些方法可以从不同角度进行分类。 1.1 从实现神经网络所使用的器件分类 1.1.1 直接基于硬件的神经网络实现。包括VLSI实现和光学实现等方法。其中VLSI实现又可分为VLSI模拟实现、数字VLSI实现及模数混合实现。 1.1.1 基于现代数字计算机的神经网络实现。这种实现方法是在现代数字计算机硬件和软件技术(尤其是并行处理技术)基础上发展起来的。每个神经元不再对应一个物理处理单元,而是对应于一个虚拟
1, the realization method of neural network There are a variety of ways to achieve neural network, these methods can be classified from different angles. 1.1 From the implementation of neural network devices used in the classification 1.1.1 Direct hardware-based neural network. Including VLSI implementation and optical implementation methods. Among them VLSI realization can be divided into VLSI simulation implementation, digital VLSI realization and module realization. 1.1.1 Based on the modern digital computer neural network. This method of implementation is based on modern digital computer hardware and software technology (especially parallel processing technology) developed. Each neuron no longer corresponds to a physical processing unit, but corresponds to a virtual