基于NARA模型和筛选方法的并行神经网络体系结构

A PARALLEL NEURAL NETWORK ARCHITECTURE BASED ON NARA MODEL AND SIEVING METHOD

本文将介绍一种并行的神经网络体系结构（ＰＮＮ），它是以ＮＡＲＡ模型和筛选方法为基础的．ＰＮＮ由一个控制网络ＣＮ和一组识别网络ＲＮｉ（ｉ＝１，２，３，…，ｐ）组成，它能够自动地将复杂问题分解为简单问题，容易实现追加学习，并且可以分析其内部状态，其结构也是模块化结构，易于硬件电路实现，可以作为一种计算机运算部件，而且ＰＮＮ具有较高的运行效率．

A parallel neural network architecture (PNN) based on NARA model and sieving method is presented in this paper. PNN is composed of a control network (CN) and a series of recognition networks (RNi, i=1,2,…, p). It has several attractive properties such as automatic decomposition of learning tasks, easy implementation of additional learning and easy analysis of internal states. PNN has modular structure and each model of it can be used as a computing component. So it can be implemented with hardware easily and work with high efficiency.