正在加载图片...
第5期 马楠等:一种模糊认知图分类器构造方法 ·595· 5%时,本文模型的分类优势逐渐体现,优于SVM、 fault management in networks using fuzzy cognitive maps /Pro- C4.5和BP的次数分别是2、3和2次,SVM优于 ceedings of International Conference Communications Converging Technologies for Tomorrow's Application.New York,1996:1558 C4.5和BP的次数分别是2次和3次,C4.5和BP [8]Stach W,Kurgan L.Modeling software development project using 分类性能相当;当噪声测度为10%时,本文模型的 fuzzy cognitive maps /Proceedings of 4th ASERC Workshop on 分类性能优于SVM、C4.5和BP的次数分别是2 Quantitative and Soft Softcare Engineering.Banff,2004:55 次,4次和4次,SVM优于C4.5和BP的次数分别 9] Dickerson J A,Kosko B.Adaptive fuzzy cognitive maps in virtual 是2次和4次,C4.5优于BP1次;当噪声测度为 worlds /World Congress on Neural Networks (WCNN 94).San 15%时,本文模型的分类优势进一步体现,优于 Diego,1994:471 [10]Kakolyris A,Stylios G,Georgopoulos V C.Fuzzy cognitive maps SVM、C4.5和BP的次数分别是5、7和6次,SVM优 application for web mining /Proceedings of the Joint 4th Confer- 于C4.5和BP的次数分别是6次和3次,C4.5优于 ence of the European Society for Fuzzy Logic and Technology.Bar- BP1次.表3~表5中每个数字单元格内容的含义 celona,2005:593 同表2.综合表2~表5可知,本文模型的在总体性 [1]Stula M,Stipanicev D,Bodrozic L.Intelligent modeling with 能上优于传统的分类器,尤其随着噪声测度增大,它 agent-based fuzzy cognitive map.Int Intell Syst,2010,25 (10):981 优越性愈加明显. [12]Peng Z,Yang B R,Liu C M,et al.Research on one fuzzy cog- 4结论 nitive map classifier.Appl Res Comput,2009,26(5):1757 (彭珍,杨炳儒,刘春梅,等.一种模糊认知图分类器的研究 模糊认知图已成为数据挖掘领域的热点问题, 计算机应用研究,2009,26(5):1757) 将它运用到分类问题不但具有深远的理论意义,而 [13]Zhang G Y,Liu Y,Wang YY.Reference algorithm of text cate- gorization based on fuzzy cognitive maps.Comput Eng Appl, 且具有广阔的应用前景.本文从系统模型的视角, 2007,43(12):155 构造了一种新的模糊认知图分类器,从模型的构建 (张桂芸,刘洋,王元元.基于模糊认知图的文本分类推理算 流程、激活函数、推理规则和学习方法等各部件进行 法.计算机工程与应用,2007,43(12):155) 了深入分析和详细阐述.本文提出的分类器在分类 [14]Wang J Z,Xing Y P,Shi P,et al.Using fuzzy cognitive map to 过程中,改善与提高了模糊认知图内涵的分类功能, effectively classify E-documents and application.Lect Notes Com- put Sei,2005,3795:591 在很大程度上解决了模糊认知图参与挖掘过程的问 [15]Zhu D Y,Mendis B S,Gedeon T.A hybrid fuzzy approach for 题.实验结果表明本文方法是有效的.集成分类器 human eye gaze pattem recognition//Proceedings of Internation- 是提高分类精度的有效方案四,模糊认知图分类器 al Conference on Neural Information Processing of the Asia-Pacific 模型的集成策略是下一步的研究重点. Neural Netcork Assembly.Berlin,2008:655 [16]Blake C,Keogh E,Merz C J.UCI repository of machine leaming 参考文献 databases [DB/OL][2011-05-20]www.ics.uci.edu/~ml- Kosko B.Fuzzy cognitive maps.Int J Man Mach Stud,1986,24: earn/MLRepository.html 65 [17]Brodley C E,Friedl M A.Identifying and eliminating mislabeled Axelrod R.Structure of Decision:the Cognitire Maps of Political training instances /Proceedings of 13th National Conference on Elites.New Jersey:Princeton University Press,1976 Artificial Intelligence.Alon Levy,1996:799 Lin C M.Model Method and Application Study of Fuzy Cognitire [18]Zhang X G.Introduction to statistical leaming theory and support Map [Dissertation].Shanghai:Donghua University,2006 vector machines.Acta Autom Sin,2000,26(1):32 (林春梅.模糊认知图模型方法及其应用研究[学位论文].上 (张学工.关于统计学习理论与支持向量机.自动化学报, 海:东华大学,2006) 2000,26(1):32) 4]Stach W,Kurgan L,Pedrycz W,et al.Genetic leaming of fuzzy 19] Dietterich TG.An experimental comparison of three methods for cognitive maps.Fuzzy Sets Syst,2005,153(3):371 constructing ensembles of decision trees:bagging,boosting,and [5]Georgopoulos V C,Malandraki G A,Stylios C D.A fuzzy cogni- randomization.Mach Learn,2000,40(2):139 tive map approach to differential diagnosis of specific language im- 120]Zhai Y,Yang B R,Qu W,et al.Study on source of classifica- pairment.Artif Intell Med,2003,29 (3):261 tion in imbalanced datasets based on new ensemble classifier. [6]Styblinski M A,Meyer B D.Signal flow graphs vs fuzzy cognitive Syst Eng Electron,2011,33 (1)196 maps in application to qualitative circuit analysis.Int I Man Mach (程云,杨炳儒,曲武,等.基于新型集成分类器的非平衡数 Stud,1991,35(2):175 据分类关键问题研究.系统工程与电子技术,2011,33(1): Ndousse TD,Okuda T.Computational intelligence for distributed 196)第 5 期 马 楠等: 一种模糊认知图分类器构造方法 5% 时,本文模型的分类优势逐渐体现,优于 SVM、 C4. 5 和 BP 的次数分别是 2、3 和 2 次,SVM 优于 C4. 5 和 BP 的次数分别是 2 次和 3 次,C4. 5 和 BP 分类性能相当; 当噪声测度为 10% 时,本文模型的 分类性能优于 SVM、C4. 5 和 BP 的 次 数 分 别 是 2 次,4 次和 4 次,SVM 优于 C4. 5 和 BP 的次数分别 是 2 次和 4 次,C4. 5 优于 BP 1 次; 当噪声测度为 15% 时,本文模型的分类优势进一步体现,优 于 SVM、C4. 5 和 BP 的次数分别是 5、7 和 6 次,SVM 优 于 C4. 5 和 BP 的次数分别是 6 次和 3 次,C4. 5 优于 BP 1 次. 表 3 ~ 表 5 中每个数字单元格内容的含义 同表 2. 综合表 2 ~ 表 5 可知,本文模型的在总体性 能上优于传统的分类器,尤其随着噪声测度增大,它 优越性愈加明显. 4 结论 模糊认知图已成为数据挖掘领域的热点问题, 将它运用到分类问题不但具有深远的理论意义,而 且具有广阔的应用前景. 本文从系统模型的视角, 构造了一种新的模糊认知图分类器,从模型的构建 流程、激活函数、推理规则和学习方法等各部件进行 了深入分析和详细阐述. 本文提出的分类器在分类 过程中,改善与提高了模糊认知图内涵的分类功能, 在很大程度上解决了模糊认知图参与挖掘过程的问 题. 实验结果表明本文方法是有效的. 集成分类器 是提高分类精度的有效方案[20],模糊认知图分类器 模型的集成策略是下一步的研究重点. 参 考 文 献 [1] Kosko B. Fuzzy cognitive maps. Int J Man Mach Stud,1986,24: 65 [2] Axelrod R. Structure of Decision: the Cognitive Maps of Political Elites. New Jersey: Princeton University Press,1976 [3] Lin C M. Model Method and Application Study of Fuzzy Cognitive Map [Dissertation]. Shanghai: Donghua University,2006 ( 林春梅. 模糊认知图模型方法及其应用研究[学位论文]. 上 海: 东华大学,2006) [4] Stach W,Kurgan L,Pedrycz W,et al. Genetic learning of fuzzy cognitive maps. Fuzzy Sets Syst,2005,153( 3) : 371 [5] Georgopoulos V C,Malandraki G A,Stylios C D. A fuzzy cogni￾tive map approach to differential diagnosis of specific language im￾pairment. Artif Intell Med,2003,29( 3) : 261 [6] Styblinski M A,Meyer B D. Signal flow graphs vs fuzzy cognitive maps in application to qualitative circuit analysis. Int J Man Mach Stud,1991,35( 2) : 175 [7] Ndousse T D,Okuda T. Computational intelligence for distributed fault management in networks using fuzzy cognitive maps / / Pro￾ceedings of International Conference Communications Converging Technologies for Tomorrow's Application. New York,1996: 1558 [8] Stach W,Kurgan L. Modeling software development project using fuzzy cognitive maps / / Proceedings of 4th ASERC Workshop on Quantitative and Soft Software Engineering. Banff,2004: 55 [9] Dickerson J A,Kosko B. Adaptive fuzzy cognitive maps in virtual worlds / / World Congress on Neural Networks ( WCNN 94) . San Diego,1994: 471 [10] Kakolyris A,Stylios G,Georgopoulos V C. Fuzzy cognitive maps application for web mining / / Proceedings of the Joint 4th Confer￾ence of the European Society for Fuzzy Logic and Technology. Bar￾celona,2005: 593 [11] Stula M,Stipanicev D,Bodrozic L. Intelligent modeling with agent-based fuzzy cognitive map. Int J Intell Syst,2010,25 ( 10) : 981 [12] Peng Z,Yang B R,Liu C M,et al. Research on one fuzzy cog￾nitive map classifier. Appl Res Comput,2009,26( 5) : 1757 ( 彭珍,杨炳儒,刘春梅,等. 一种模糊认知图分类器的研究. 计算机应用研究,2009,26( 5) : 1757) [13] Zhang G Y,Liu Y,Wang Y Y. Reference algorithm of text cate￾gorization based on fuzzy cognitive maps. Comput Eng Appl, 2007,43( 12) : 155 ( 张桂芸,刘洋,王元元. 基于模糊认知图的文本分类推理算 法. 计算机工程与应用,2007,43( 12) : 155) [14] Wang J Z,Xing Y P,Shi P,et al. Using fuzzy cognitive map to effectively classify E-documents and application. Lect Notes Com￾put Sci,2005,3795: 591 [15] Zhu D Y,Mendis B S,Gedeon T. A hybrid fuzzy approach for human eye gaze pattern recognition / / Proceedings of Internation￾al Conference on Neural Information Processing of the Asia-Pacific Neural Network Assembly. Berlin,2008: 655 [16] Blake C,Keogh E,Merz C J. UCI repository of machine learning databases[DB /OL][2011--05--20] www. ics. uci. edu / ~ ml￾earn /MLRepository. html [17] Brodley C E,Friedl M A. Identifying and eliminating mislabeled training instances / / Proceedings of 13th National Conference on Artificial Intelligence. Alon Levy,1996: 799 [18] Zhang X G. Introduction to statistical learning theory and support vector machines. Acta Autom Sin,2000,26( 1) : 32 ( 张学工. 关于统计学习理论与支持向量机. 自动化学报, 2000,26( 1) : 32) [19] Dietterich T G. An experimental comparison of three methods for constructing ensembles of decision trees: bagging,boosting,and randomization. Mach Learn,2000,40( 2) : 139 [20] Zhai Y,Yang B R,Qu W,et al. Study on source of classifica￾tion in imbalanced datasets based on new ensemble classifier. Syst Eng Electron,2011,33( 1) : 196 ( 翟云,杨炳儒,曲武,等. 基于新型集成分类器的非平衡数 据分类关键问题研究. 系统工程与电子技术,2011,33 ( 1) : 196) ·595·
<<向上翻页
©2008-现在 cucdc.com 高等教育资讯网 版权所有