一种粒子群优化原型模式修正力度的协同分类方法

英文篇名：A Synergetic classification algorithm based on prototype modify with particle swarm optimization measure
中文刊名：信号处理
英文刊名：Signal Processing
作者：邹刚 ; 姚伟 ; 敖永红 ; 孙即祥 ; 陈森林
英文作者：ZOU Gang~(1 ; 2) YAO Wei~1 AO Yong-hong~2 SUN Ji-xiang~1 CHEN Sen-lin~3 (1.Institute of electronic engineering ; National university of Defence Technology ; ChangSha China 410073 ; 2.Center of information ; National university of Defence Technology ; ChangSha China 410073 ; 3.China Hunan Tumor Hospital ; Changsha ; China ; 410084)
中文关键词：协同神经网络 ; 原型模式重构 ; 粒子群优化
英文关键词：Synergetic Neural Network(SNN) ; prototype modify ; particle swarm optimization(PSO)
出版日期：2010-04-25
机构：国防科技大学电子工程学院;国防科技大学信息中心;湖南省肿瘤医院病理科;
年：2010
期：04
出版单位：信号处理

摘要

协同模式识别是一种有着抗噪声、抗缺损、强鲁棒性等诸多优良特性的模式识别方法,其中原型模式的选取对模式识别结果有着决定性的作用,其选取直接决定着模式识别的结果和效果,各种方法中信息反馈修正的方法能获得较好的效果,但易出现信息饱和的问题;提出了一种粒子群优化修正力度的处理机制,能有效改善此问题,获得最优原型;将改进的算法应用于纹理和鼻咽癌细胞图像识别,结果表明,该方法能有效地提高协同神经网络的识别率和可靠性,且识别速度也有提高。
The synergetic pattern recognition is a new way of pattern recognition with many excellent features such as noise resistance, deformity resistance,and better robustness,the selection of prototype patterns is very important to pattern recognition of synergetic approach,which set the tone for the recognition performance of synergetic approach,the superposition modify of information is better in the existing methods of prototype selection,prototype modify method with particle swarm optimization measure is applied to avoided information saturation,and get the optimal prototype.experiment result on Brodatz texture images and nasopharyngeal carcinoma cell images shows that the new algorithm can effectively search the optimal prototype patterns,the synergetic recognition method proposed in this paper is more available than classical synergetic pattern recognition method,and excellent,correct and fast recognition result has been achieved,with good potential clinical application.

引文

[1]Hinton G E.Connectionist learning procedures.Artificial Intelligence,1989,40(1-3 ):185-234.
    [2]刘秉瀚,王伟智,方秀端.协同模式识别方法综述[J].系统工程与电子技术,2003.25(6):758-762.
    [3]Shutton R S.Two problems with backpropagation and other steepest-decent learning procedures for networks Proceedings of 8th Annual Conference of the Cognitive Science, Erlbaum,Hillsdale,NJ,1986:823-831.
    [4]Haken H.协同计算机和认知——神经网络的自上而下方法[M].杨家本,译,北京:中国计算机学会学术著作丛书,清华大学出版社,1994.
    [5]Wagner T,Bobel F G.Testing Synergetic Algorithms with Industrial Classification Problems[J].Neural Networks. 1994,17(8):1313-1321.
    [6]董火明,高隽,陈定国,陈迎春.协同神经网络聚类型学习算法.合肥工业大学学报(自然科学版J) 2002.25(4):493-494.
    [7]韩绍卿,李夕海,宋仔标,刘代志.基于模糊C-均值的原型模式选择及其在核爆地震识别中的应用.核电子学与探测技术[J]2007,27(5):820-821.
    [8]王海龙,戚飞虎,詹劲峰基于遗传算法的原型模式选取算法.计算机工程[J],2000,26(10):19-20.
    [9]王海龙,戚飞虎.基于信息叠加的学习算法.红外与毫米波学报[J].2000,19(3):26-27.
    [10]田晓东,刘忠.改进原型模式重构的协同识别算法及应用.青岛大学学报[J]2005,20(3):86-87;
    [11]Ma X L and Jiao L C.An effective learning algorithm of synergetic neural network.Lecture Notes in Computer Science. 2004,3173:258-263.
    [12]缑水平,焦李成,田小林.基于免疫克隆聚类协同神经网络的图像识别电子与信息学报[J].2008,30(2): 264-265.
    [13]刘秉瀚,王伟智,郑智勇.基于竞争最优原型的淋巴细胞协同分类.中国体视学与图像分析[J].2006,11 (2):71-74.
    [14]Fang Xiuduan,Liu Binghan,Wang Weizhi.An Improved Prototype Pat-tern Selection Algorithm[J].Computer Science,2002,29(9):134-136.
    [15]马正中,阚秀,刘树范.诊断细胞病理学河南科学技术出版社2000.5第一版.
    [16]王伟智,刘秉瀚,朱敏琛.基于协同方法交通状态识别中国体视学与图像分析[J]2007,12(1):38-39.
    [17]Haken H.Synergetic Computers and Recognition-a-topdown Approach to Neural Nets.Berlin:Springer-Verlag, 1991:45-80.
    [18]欧阳森,宋政湘,王建华等.一种快速收敛的遗传算法计算机应用研究[J]2003(9):51-52.