深度学习目标识别算法及运用研究  

Deep Learning Target Recognition Algorithm and Application Research

作　　者：肖立 潘天恒 吴晓凌[1] Xiao Li;Pan Tianheng;Wu Xiaoling

机构地区：[1]武汉生物工程学院,湖北武汉430010

出　　处：《数码设计》2018年第16期23-23,共1页Peak Data Science

摘　　要：文章主要论述了深度学习目标识别的算法,对其主要运用进行了研究分析。神经网络的结构是一种基于人脑结构设计的技术手段,卷积神经网络属于一种特殊的深度前馈网络信息,为了避免在不同层级之间信息链接产生参数冗余,通过局部链接以及权值共享的方式进行处理,卷积神经网络具有稀疏特征,可以链接符合生物神经元稀疏的响应特性,可以有效的降低网络的参数规模,使得模型训练更为简单便捷。this paper mainly discusses the algorithm of deep learning target recognition and analyzes its main application.The structure of the neural network is a technical means based on the design of the structure of the human brain.The convolution neural network belongs to a special deep feedforward neUvork information.In order to avoid parameter redundancy between information links at different levels,through local links and weight sharing.The way to process.The convolution neural network has sparse characteristics,wliich can be linked to the sparse response characteristics of biological neurons,and can effectively reduce the parameter scale of the network,making the model training more simple and convenient.

关 键 词：深度学习目标识别算法 运用研究 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]