簡易檢索 / 詳目顯示

回結果列表
研究生： 吳柏言
Wu, Po-Yen
論文名稱： 針對兩類及多類分類問題之相關學習策略
Some Learning Procedures for Binary and Multiple Classification Problems
指導教授： 陳瑞彬
Chen, Ray-Bing
學位類別： 碩士
Master
系所名稱： 管理學院 - 統計學系
Department of Statistics
論文出版年： 2015
畢業學年度： 103
語文別： 英文
論文頁數： 49
中文關鍵詞： 兩類分類器 、線性區別分析 、線性迴歸 、巨量資料 、主動式學習 、AUC
外文關鍵詞： binary classifier, linear discriminant analysis, linear regression, big data, active learning, AUC
相關次數： 點閱：151  下載：15 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報

    • 將測試樣本投影至適合的維度(PuLSIF) 以及若單純投影可能無法得到最佳解，多經過旋轉基底這步驟(PuLSIF_RD)，上述的兩種方法是為了改進uLSIF；而我們也運用一樣的想法在處理多分類問題的LSPC。不過改進並不顯著，甚至所花費的時間卻更多。
    另外，在類別個數等於2 時，線性迴歸與費雪的線性區別分析有所關聯。我們可以想像如果樣本投影至線性迴歸所建立的迴歸線並藉由這個類別分數去分配那些樣本。
    萬一訓練樣本集成本過高或是訓練樣本數過大這樣的情況，這樣的主動式學習情況，使用序列方法是自然的作法；也就是說我們利用迭代的方法去減少訓練樣本集的使用。
    因此在此篇論文呈現了利用迴歸區別分類器搭配一些準則來解決節省訓練集樣本之成本。再者，我們重點不著重在配適的迴歸模型正確與否，而是分類結果如何。只要利用小樣本的分類結果與全數的訓練集樣本的結果不會差太多，即解決了目的，甚至這樣的結果也與巨量資料有關。而數值結果在分類正確率及AUC 都反映了這樣的結果。

    The projection uLSIF (PuLSIF) and projection uLSIF rotation data(PuLSIF_RD) are the improvements
    of unconstrained least-square importance fitting (uLSIF). We apply these ideas
    which try to the projection and rotation subspace to improve the least-square probabilistic
    classification (LSPC). However, the improvement is not significant and time consuming is
    much longer.
    Moreover, we are informd that the naive linear regression method can be linked the
    Fisher’s discriminant analysis when binary classes. We imagine that samples are projected
    along the line constructed by the linear regression method and allocate the samples. If such
    cases as the training sample are expensive or the sample size are extremely large are encountered.
    And under this active learning scenario, the sequential method is a nature technique.
    In short, we use the iteration to reduce the training sample size.
    Thus in this paper, there will be presented some criteria with the regression discriminant
    classifier and these criteria can economize on training samples. Besides, we do not focus
    on the modeling part but the predition part. As long as two performances of small training
    sample and all training sample size are close, then solve the problem about training cost.
    Indeed, its result can be involved in big data issue. Numerical result can conclude that these
    strategies are comparable to the all sample size into regression discriminant classifier in accuracy,
    even AUC part.

    摘要 I Abstract II Acknowledgements III Contents IV List of Tables VI 1 Introduction 1 2 The Improvement of LSPC 3 2.1 Framework of LSPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.1.1 Two models with the basis function . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Methodology for PRLSPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.2.1 The idea of PRLSPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.2.2 The PRLSPC setting and procedure . . . . . . . . . . . . . . . . . . . . . . 7 2.3 Results of PRLSPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.3.1 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.3.2 Simulation Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3.3 Results for Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Appendix 2.A Others Results of PRLSPC . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3 Sequential Procedure in the RDC 21 3.1 The Methodology of RDC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.2 Sequential Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2.1 Next Sample Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2.2 Stopping Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4 Results of Sequential Procedure in the RDC 27 4.1 Simulation Data and Numerical Result . . . . . . . . . . . . . . . . . . . . . . . . 27 4.1.1 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.1.2 Simulation Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.1.3 Numerical Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4.1.4 Real Data Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.1.5 Idea Attemption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5 Conclusion and Future Work 39 5.1 Conclusion for Sequential Procedure in the RDC . . . . . . . . . . . . . . . . . . . 39 5.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 References 41 Appendix A Result of the Higher Dimensional Data 43

    Chen, C. C. (2014). A classification approach based on density ratio estimation with subspace
    projection. master thesis, Department of Statistics, National Cheng KungUniversity,
    Taiwan.
    Erdogmus, D., Rao, Y. N., Principe, J. C., Zhao, J., and Hild II, K. E. (2002). Simultaneous
    extraction of principal components using givens rotations and output variances. in Proc.
    ICASSP, pages 1069–1072.
    Flake, G. W. and Lawrence, S. (2002). Efficient svm regression training with smo. Machine
    Learning, 46:271–290.
    Hastie, T., Tibshirani, R., and Buja, A. (1994). Flexible discriminant analysis by optimal scoring.
    J. Amer. Statist. Assoc., 89:1255–1270.
    Ho, T. K. and Kleinberg, E. M. (1996). Building projectable classifiers of arbitrary complexity.
    In Proceedings of the 13th International Conference on Pattern Recognition, pages 880–885.
    Kanamori, T., Hido, S., and Sugiyama, M. (2009). A least-squares approach to direct importance
    estimation. Journal ofMachine Learning Research, 10:1391–1445.
    Lai, T. L., Robbins, H., andWei, C. Z. (1979). Strong consistency of least squares estimates in
    multiple regression ii. J.Multivariate Anal, 9:346–361.
    Lai, T. L. and Wei, C. Z. (1982). Least squares estimates in stochastic regression models with
    applications to identification and control of dynamic systems. Annals of Statistics, 10:154–
    166.
    Ramana, B. V., Babu, M. S. P., and Venkateswarlu,N. B. (2011). A critical study of selected classification
    algorithms for liver disease diagnosis. International Journal of Database Management
    Systems, 3(2):506–516.
    Ramana, B. V., Babu, M. S. P., and Venkateswarlu, N. B. (2012). A critical comparative study of
    liver patients from usa and india: An exploratory analysis. International Journal of Computer
    Science Issues, 9(2):506–516.
    Rätsch, G., Onoda, T., and Müller, K.-R. (2001). Soft margins for adaboost. Machine Learning,
    42:287–320.
    Roea, B. P., Yang, H. J., Zhu, J., Liu, Y., Stancuc, I., andMcGregor, G. (2005). Boosted decision
    trees as an alternative to artificial neural networks for particle identification. Nucl. Instrum.
    Meth, A543:577.
    Sugiyama, M. (2010). Superfast-trainable multi-class probabilistic classifier by least-squares
    posterior fitting. IEICE Transactions on Information and Systems, E93–D(10):2690–2701.
    Wang, Z. F. and Chang, Y. I. (2013). Sequential estimate for linear regression models with
    uncertain number of effective variables. Metrika, 76:949–978.
    Webb, A. R. (2002). Statistical Pattern Recognition. Wiley.
    Yamada, M., Sugiyama, M., Wichern, G., and Simm, J. (2011). Improving the accuracy of
    least-squares probabilistic classifiers. IEICE Transactions on Information and Systems.
    Yeh, I.-C., Yang, K.-J., and Ting, T.-M. (2009). Knowledge discovery on rfm model using
    bernoulli sequence. Expert Systems with Applications, 36:5866–5871.

    下載圖示 校內：2020-08-17公開
    校外：2020-08-17公開
    QR CODE