臺灣博碩士論文加值系統

本研究當中，提出對高維度資料進行分類時，先利用bagging隨機選取樣本然後搭配自動選取子空間維度的加權隨機子空間法。其中特徵的選擇是利用
核平滑化法在進行訓練時不斷更新特徵重要性的分配再根據分配來選取特徵。以此法所建構出之多重分類器系統能夠解決選取樣本及特徵之問題，並能避免小樣本分類不佳之情況。利用三種特徵分配加權方式，原始隨機子空間的均一分布加權法、分類正確率分布加權法及線性區別分析分離量分布加權法針對bagging以及加權隨機子空間法加以改善。實驗結果在高維度資料當中顯示所提的方法效能比bagging及加權隨機子空間法較佳。

In this paper, using bagging method to select data randomly and a weighted random subspace method with automatic subspace dimensionality selection have been proposed for classifying high dimensional data. The dimensionality selection method is based on the importance distribution of dimensionality estimated by kernel smoothing technique during the algorithm training. The multiple classifier system can solve the problem of selecting samples and features and prevent that the poor performance of classifying small sample size. Three feature weighting methods based on uniform distribution in original random subspace, normalized re-substitution accuracy and Fisher’s Linear Discriminate Analysis separability are introduced for improving the bagging method and weighted random subspace method. Experimental result shows that the proposed algorithm outperforms the bagging method and weighted random subspace method in high dimensional data.

第一章 緒論 1
第一節 研究動機 1
第二節 研究目的 3
第三節 符號說明 3
第二章 文獻探討 6
第一節 特徵選取 6
第二節 特徵萃取 6
第三節 分類器 8
壹、高斯分類器 8
貳、k最近鄰分類器 9
參、支撐向量機分類器 9
第四節 多重分類器與結合策略 11
第五節 Bagging 14
第六節 Random Subspace Method 14
第七節 Weighted Random Subspace Method 16
第三章 Bagging及WRSM之結合 21
第一節 研究概念 21
第二節 演算法 21
壹、BG-RSM-KS演算法 21
貳、BG-WRSM-KS1演算法 22
參、BG-WRSM-KS2演算法 23
第三節 資料描述 24
壹、Washington DC Mall資料集 24
貳、Indian Pine Site資料集 26
參、教育測驗資料集 27
第四章 研究結果及討論 30
第五章 結論和未來發展 62
參考文獻 64

中文部份
郭伯臣、吳慧珉、楊晉民、柯立偉、白家豪（民92）。樣式辨識技術於學生補救教學分組之應用－以國小數學領域「扇形」單元為例，九十二學年度師範學院教育學術論文發表會，台南師院，10月24-25日。
郭伯臣（民95）統計樣式辨認於測驗資料之應用。測驗統計年刊，第十三輯下期，pp.146-170。

英文部份
Bertoni, A., Folgieri, R. & Valentini, G. (2004). “Random Subspace Ensembles for the Bio-molecular Diagnosis of Tumors,” Models and Metaphors from Biology to Bioinformatics Tools, NETTAB.
Bertoni, A., Folgieri, R. & Valentini, G. (2004). “Feature Selection Combined with Random Subspace Ensemble for Gene Expression Based Diagnosis of Malignancies,” Workshop on Neural Networks, Methods for bioinformatics and biostatistics.
Breiman, L. (1996). Bagging Predictors, Machine Learning, 24(2), 123-140.Freund,
Y., Schapire, RE., Experiments with A New Boosting Algorithm.
Dietterich, T. G. (2000). Ensemble Methods in Machine Learning, First International
Workshop of Multiple Classifier Systems, vol. 1857, pp. 1-15
Duin, R. P. W. (2003). A Matlab Toolbox for Pattern Recognition, Software available
at http://ftp.ph.tn.tudelft.nl/pub/bob/prtools/.
Fisher, R.A. (1936). The use of multiple measures in taxonomic problems. Ann.
Eugenics, vol. 7, pp. 179-188
Fukunaga, K. (1990). Introduction to Statistical Pattern Recognition, San Diego:
Academic Press Inc.
Ho, T. K. (1998). “The Random Subspace Method for Constructing Decision
Forests, ”Pattern Analysis and Machine Intelligence, IEEE Trans. Vol. 20, no. 8, pp.832-344.
Ho, T. K. (1998). “Nearest Neighbors in Random Subspaces,” Proceedings of the
Second International Workshop on Statistical Techniques in Pattern Recognition,
Sydney, Australia, August 11-13, pp. 640-648.
Kittler, J., Hatef, M., Duin, R.P.W., & Matas, J. (1998). On Combining Classifiers,
IEEE Transactions on Pattern Analysis and Machine Intelligence, 20(3),
226-239.
Kuo, B-C , Hsieh Y-C , Liu H-C, & Chao R-M (2005). “A Random Subspace Method with Automatic Dimensionality Selection for Hyperspectral Image Classification, ”Proceedings of International Geoscience and Remote Sensing
Symposium.
Kuo, B-C. & Landgrebe, D. A. (2001). Improved Statistics Estimation and Feature
Extraction for Hyperspectral Data Classification, Technical Report, Purdue
University, West Lafayette, IN., TR-ECE 01-6.
Kuo, B-C., Pai, C-H., Sheu, T-W., & Chen, G-S. (2004). “Hyperspectral Data Classification Using Classifier Overproduction and Fusion Strategies, ”Proceedings of International Geoscience and Remote Sensing Symposium.
Landgrebe, D. A. (2003). Signal Theory Methods in Multispectral Remote Sensing, John Wiley and Sons, Hoboken, NJ: Chichester.
Roli, F., Giacinto, G.(2002). Design of Multiple Classifier Systems, H. Bunke and
A.Kandel (Eds.) Hybrid Methods in Pattern Recognition. Skurichina, M. & Duin, R. P. W. (2001). “Bagging and the Random Subspace Method for Redundant Feature Spaces,” Second International Workshop of Multiple Classifier Systems, pp.1-10.
Skurichina, M. & Duin, R. P. W. (2002). “Bagging, Boosting and the Random Subspace Method for Linear Classifiers,” Pattern Analysis & Applications, Vol. 5, no. 8, pp.121-135.
Skurichina, M. & Duin, R.P.W.(1998). Bagging for Linear Classifiers, Pattern Recognition, 31(7), 909-930.
Skurichina, M., Duin, RPW., (2000). The Role of Combining Rules in Bagging and Boosting, Lecture Notes in Computer Science, 1876, 631-640.