OUTLINE Introduction Fuzzy cMeans Algorithm FCM Feature Selected
- Slides: 41
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
数据集的C划分 Crisp c-Partition Fuzzy c-Partition
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
特征选择性FCM算法 FCM目标函数： 基于特征加权的 FCM目标函数： Jie Li, Xinbo Gao, Licheng Jiao, “A novel feature weighted fuzzy clustering algorithm”, D. Slezak et al. (Eds. ): RSFDGr. C 2005, LNAI 3641, pp. 412 -420, 2005, Springer-Verlag Berlin Heidelberg 2005
Relief. F算法 Ø Relief算法是Kira和Rendell在 1992年提出的，限于解 决两类的分类问题的特征选择； Ø 1994年Kononenko扩展了Relief算法，使得Relief. F可 以解决多类问题的特征选择； Ø Relief. F算法是给特征集中每一特征赋予一定的权重。 Kira K. , Rendell L A. , A practical approach to feature selection, Proceedings of the 9 th International Workshop on Machine Leaning, San Francisco, CA: Morgan Kaufmann, 1992, 249 -256
实验结果 IRIS数据由四维空间中的150个样本点组成，每一个样本的4个分量分别表示 IRIS的Petal Length，Petal Width，Sepal Length和Sepal Width。 包含了3个IRIS种类Setosa，Versicolor和Virginica，每类各有50个样本。 其中Setosa与其它两类间较好地分离，而Versicolor和Virginica之间存在交迭。
W-k-Means Algorithm Joshua Zhexue Huang，Michael K. Ng, Hongqiang Rong, and Zichen Li, Automated Variable Weighting in k-Means Type Clustering, IEEE Trans. on PAMI, 27(5): 657 -668, 2005
W-k-Means Algorithm Joshua Zhexue Huang，Michael K. Ng, Hongqiang Rong, and Zichen Li, Automated Variable Weighting in k-Means Type Clustering, IEEE Trans. on PAMI, 27(5): 657 -668, 2005
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
Sample selective FCM 目标函数： 迭 代 公 式
FCM for Large Data Set 原子聚类 典型样本 样本加权 Jie Li, Xinbo Gao, Licheng Jiao, “A Novel Typical-Sample-Weighted Clustering Algorithm for Large Data Sets ”, Lecture Notes in Artificial Intelligence, LNAI 3801: 696 -703, 2005
FCM for Large Data Set 对聚类中 心的影响 算法的可 扩展性 Jie Li, Xinbo Gao, Licheng Jiao, “A Novel Typical-Sample-Weighted Clustering Algorithm for Large Data Sets ”, Lecture Notes in Artificial Intelligence, LNAI 3801: 696 -703, 2005
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
OUTLINE Introduction Fuzzy c-Means Algorithm (FCM) Feature Selected FCM Algorithm Sample Selected FCM Algorithm Selective Ensemble Concluding Remarks References
References 1. Jie Li, Xinbo Gao, Licheng Jiao, A novel feature weighted fuzzy clustering algorithm, D. Slezak et al. (Eds. ): RSFDGr. C 2005, LNAI 3641, pp. 412 -420, 2005, Springer-Verlag Berlin Heidelberg 2005 2. Kira K. , Rendell L A. , A practical approach to feature selection, Proceedings of the 9 th International Workshop on Machine Leaning, San Francisco, CA: Morgan Kaufmann, 1992, 249 -256 3. Joshua Zhexue. Huang，Michael K. Ng, Hongqiang Rong, and Zichen Li, Automated Variable Weighting in k-Means Type Clustering, IEEE Trans. on PAMI, 27(5): 657 -668, 2005 4. 高新波，李�， “基于加� FCM与����指�的多���像自�分割算法 ”，� 子学� ，32(4): 661 -664, 2004 5. Jie Li, Xinbo Gao, Licheng Jiao, A Novel Typical-Sample-Weighted Clustering Algorithm for Large Data Sets, CIS 2005 6. Breiman L. Bagging predicators. Machine Learning, 1996, 24(2): 123− 140. 7. 唐�，周志�，基于 Bagging的��性聚�集成，�件学�， 16(4): 496 -502, 2005
THANK YOU Xinbo Gao School of Electronic Engineering Xidian University Xi’an 710071 P. R. China xbgao@IEEE. org http: //see. xidian. edu. cn/faculty/xbgao Xidian University, Xi’an, China © 2006