毕业论文
神经网络在模式识别中的简单分析及应用
摘 要
模式识别就是机器识别、计算机识别或者机器自动化识别,目的在于让机器自动识别事物,使机器能做以前只能由人类才能做的事,具备人所具有的对各种事物与现象进行分析、描述与判断的部分能力。它研究的目的就是利用计算机对物理对象进行分类,在错误概率最小的条件下,使识别的结果尽量与客观事物相符合。
随着人们对人工神经网络的不断地认识,神经网络是指用大量的简单计算单元构成的非线性系统,它在一定程度和层次上模仿了人脑神经系统的信息处理、存储及检索功能,因而具有学习、记忆和计算等智能处理功能。这样人们利用人工神经网络具有高度并行性高度非线性全局作用良好的容错性与联想记忆功能自适应、自学习功能MATLAB神经网络工具箱中的神经网络模型,对经过训练的神经网络可以有效地提取信号、语音、图像等感知模式的特征,并能解决现有启发式模式识别系统不能很好解决的不变量探测、抽象和概括等问题。这样神经网络可应用于模式识别的特征提取、聚类分析、边缘检测、信号增强以及噪声抑制、数据压缩等各个环节。使用机器来进行模式的识别是一项非常有用的工作,能够辨别符号等系列的机器是很有价值的。目前,模式识别技术可以应用指纹识别、IC卡技术应用、字符识别等实例。模式识别成为人工神经网络特别适宜求解的一类问题。因此,神经网络技术在模式识别中也得到广泛应用与发展。
关键词:模式识别;人工神经网络;神经网络模型;神经网络技术
Abstract
Pattern Recognition is the machine identification, computer identification or identification of machine automation, machine aimed at automatic identification of things to do before the machine can only be made by man can do, with people with all kinds of things and on an analysis of the phenomenon, described with the ability to determine the part. It is the purpose of the study of the physical object to use the computer for classification, the probability of the smallest in the wrong conditions, so that the results of recognition as far as possible in line with objective things.
As artificial neural network to recognize the continuing, neural network refers to a large number of simple calculation unit consisting of non-linear system, which to some extent and level system to imitate the human brain's information processing, storage and retrieval functions, which has learning, memory and computing functions such as intelligent processing. Such people to use artificial neural network with a high degree of parallelism, the overall role of a high degree of non-linear and good fault tolerance and associative memory function, and have good self-adaptive, self-learning function, such as prominent features, the availability of MATLAB neural network toolbox The neural network model trained neural network can effectively extract the signal, voice, video and