建立在伽瑪散度下穩健模型的調整參數之選取

Yi Hsiao; 蕭奕

請用此 Handle URI 來引用此文件： http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/1203

完整後設資料紀錄

DC 欄位	值	語言
dc.contributor.advisor	杜憶萍(I-Ping Tu)
dc.contributor.author	Yi Hsiao	en
dc.contributor.author	蕭奕	zh_TW
dc.date.accessioned	2021-05-12T09:34:11Z	-
dc.date.available	2020-07-02
dc.date.available	2021-05-12T09:34:11Z	-
dc.date.copyright	2019-07-02
dc.date.issued	2018
dc.date.submitted	2019-05-27
dc.identifier.citation	[1] Shinto Eguchi Akifumi Notsu. Robust clustering method in the presence of scattered observations. Neural Computation, 28(6):1141–1162, 2016. [2] PeterJ.Huber.Robustestimationofalocationparameter.TheAnnalsofMathematical Statistics, 35(1):73–101, 1964. [3] Michael P. Windham. Robustifying model fitting. Journal of the Royal Statistical Society. Series B (Methodological), 57(3):599–609, 1995.
dc.identifier.uri	http://tdr.lib.ntu.edu.tw/handle/123456789/1203	-
dc.description.abstract	Windham在1995年的論文Robustifying Model Fitting提出權重分布方法，解決在存有異常數據情況下可有效的估計均值，這個權重分布使用了一個參數，這個參數會影響到均值的估計表現。Windham同時提出此參數的選取方法，但我們發現在一些數值模擬例子中，這個參數選取方法表現並不佳。我們提出另一個選取方法，在數值模擬上有較佳的表現。除了一般的均值估計，我們也成功地把此方法應用在群聚分析。	zh_TW
dc.description.abstract	In 1995, Windham came out with an idea of weighted distribution in his thesis, Robustifying Model Fitting, and he used the idea to find a mean estimator when there are outliers in the original data. There is a tuning parameter in this estimator, and selecting the parameter will affect the mean estimate in the same data. In the same thesis, he also suggested a criterion of selecting the tuning parameter, but we found out that this criterion wasn’t doing well in some simulations. Considering the problem, we propose another criterion which can derive a better mean estimator. Besides, we can also apply this method to clustering problem.	en
dc.description.provenance	Made available in DSpace on 2021-05-12T09:34:11Z (GMT). No. of bitstreams: 1 ntu-107-R05246014-1.pdf: 1322277 bytes, checksum: 288717b626d751660d6e4a15006119d0 (MD5) Previous issue date: 2018	en
dc.description.tableofcontents	口試委員會審定書 iii 誌謝 v Acknowledgements vii 摘要 ix Abstract xi 1 Introduction 1.1 What is Robust?............................... 1 1.2 How to get Robust?............................. 3 1.3 Why need to do this question? ....................... 6 2 Literature Review 2.1 Normal Robust Model ........................... 9 2.2 GeneralDescription............................. 10 2.3 SelectionCriterion ............................. 12 2.4 γ-estimate and Weighted Robustified estimate . . . . . . . . . 14 3 Our Selection Criterion 3.1 gθ is univariate normal,θ=μ ....................... 17 3.2 gθ is bivariate normal,θ=μ ........................ 18 3.3 gθ is univariate normal,θ=(μ,σ2)T ................... 19 3.4 gθ is qGaussian,θ=μ ........................... 21 4 Numerical Examples 4.1 One Dimension Case ............................ 27 4.2 Two Dimension-One Component with Outliers . . . . . . . . . . 31 4.3 VarianceUnknown ............................. 32 4.4 qGaussianModel .............................. 33 Bibliography 37
dc.language.iso	en
dc.subject	穩健估計	zh_TW
dc.subject	影響函數	zh_TW
dc.subject	伽瑪散度	zh_TW
dc.subject	群聚分析	zh_TW
dc.subject	gamma-divergence	en
dc.subject	robust estimate	en
dc.subject	influence function	en
dc.subject	clustering	en
dc.title	建立在伽瑪散度下穩健模型的調整參數之選取	zh_TW
dc.title	Robust Model Fitting - Selection of Tuning Parameters in the Aspect of Gamma Clustering	en
dc.type	Thesis
dc.date.schoolyear	107-2
dc.description.degree	碩士
dc.contributor.oralexamcommittee	陳素雲(Su-Yun Huang),姚怡慶(Yi-Ching Yao),陳定立(Ting-Li Chen),陳宏(Hung Chen)
dc.subject.keyword	穩健估計,影響函數,伽瑪散度,群聚分析,	zh_TW
dc.subject.keyword	robust estimate,influence function,gamma-divergence,clustering,	en
dc.relation.page	37
dc.identifier.doi	10.6342/NTU201900783
dc.rights.note	同意授權(全球公開)
dc.date.accepted	2019-05-28
dc.contributor.author-college	理學院	zh_TW
dc.contributor.author-dept	應用數學科學研究所	zh_TW
顯示於系所單位：	應用數學科學研究所

文件中的檔案：

檔案	大小	格式
ntu-107-1.pdf	1.29 MB	Adobe PDF	檢視/開啟

顯示文件簡單紀錄

系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。