基于函数链神经网络的深度分类器  

Functional⁃Link Neural Network Based Deep Classifier

作　　者：谢润山 王士同 XIE Runshan;WANG Shitong(School of Artificial Intelligence and Computer Science,Jiangnan University,Wuxi,214122,China;Jiangsu Key Laboratory of Media Design and Software Technology(Jiangnan University),Wuxi,214122,China)

机构地区：[1]江南大学人工智能与计算机学院,无锡214122 [2]江苏省媒体设计与软件技术重点实验室(江南大学),无锡214122

出　　处：《南京航空航天大学学报》2020年第5期736-745,共10页Journal of Nanjing University of Aeronautics & Astronautics

基　　金：国家自然科学基金(61572236)资助项目。

摘　　要：目前的宽度学习系统(Broad learning system,BLS)通过所建立的一系列映射节点和增强节点来形成联合节点。因为联合节点与输出层的线性连接,网络权值可以用求解伪逆的方法快速求得,避免了耗时的训练过程,从而成为快速而高效的学习方法。然而在追求高精度结果的过程中,BLS对于增强节点数量的需求过于巨大,容易造成过拟合问题。为此,本文提出了基于函数链神经网络(Functional⁃link neural network,FLNN)的深度分类器(FLNN based deep classifier,FLNNDC),旨在提供一种更加简单却又不失精度的BLS变体结构。FLNNDC将几个轻量级的BLS子系统堆积成栈式结构,每一个轻量级的BLS子系统随机选择一部分映射节点生成增强节点,而不是全部映射节点。和原宽度结构相比,在几个主流数据集上的实验结果表明本文所提出的FLNNDC分类器具有网络结构更小且学习速度更快的优势。The existing broad learning system(BLS)forms its union nodes by generating a series of mapping nodes and enhancement nodes.Due to the linear connection between the union nodes and the output layer,the weights of the network built by BLS can be obtained quickly by using the corresponding pseudo-inverse computation,thus avoiding the time-consuming training process.As a result,BLS becomes very fast and efficient.However,in order to achieve satisfactory performance,BLS quite often needs too many enhancement nodes,which may cause over-fitting phenomenon.A functional-link neural network(FLNN)based deep classifier,called FLNNDC,is proposed to circumvent the above drawback.FLNNDC stacks several lightweight BLS sub-systems into a stacked structure,while each lightweight BLS sub-system is built by generating enhancement nodes from randomly selected mapped features instead of all the mapped features.Experimental results on several popular datasets demonstrate the effectiveness of FLNNDC in the sense of both downsized structure and less running time.

关 键 词：函数链神经网络 宽度学习 栈式结构 深度学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]