English  |  正體中文  |  简体中文  |  Items with full text/Total items : 26988/38789
Visitors : 2350964      Online Users : 33
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/11203

Title: 類神經網路應用於基因資訊之同步叢集分析暨驗證
Applying Neural Networks to Current Clustering/Validation for Gene Expression Data
Authors: 王榮華
Contributors: NTOU:Department of Electrical Engineering
國立臺灣海洋大學:電機工程學系
Keywords: 類神經網路;叢集分析;基因資訊;DNA微陣列;非對稱式引力同步聚合
shrinking transformation;extended adjacency;feedforward neuralnetworks;minimum Spanning Tree;microarray
Date: 2005
Issue Date: 2011-06-28T08:08:38Z
Publisher: 行政院國家科學委員會
Abstract: 摘要:本計畫提出一新型叢集分析方法以應用於基因資訊之處理,該方法係藉結合一新型類神經網路架構(非對稱式引力同步聚合網路,AGAN)之平行運算學習能力以及統計分析方法,期獲致同時進行基因資訊之叢集分析暨驗證(Clustering and Validation)之功效,能大幅改進運算速度及叢集可信度。 隨著生物科技的快速發展,基因序列資訊及實驗處理過程與條件控制等相關資料亦快速累積,因此亟需研發高效率的資料處理及分析技術,其能自大量基因表達序列與微陣列基因表達資料庫中,藉叢集分析等方法粹取訊息以找出契合目標生物特性的基因,以配合生命科學、數學、統計等領域的研發成果,乃成為後基因組時代(Post-Genomic Era)資訊計算研究的新課題。然而鑑於以往習知叢集分析方法(如SOM )須先給定分群數目,但最佳的分群數目k卻是未知,導致叢集可信度無法確立。即便預知最佳k值,卻往往受限於隨機給定之初始狀態initialization 而收斂至不正確之local minimum。因此本計畫提出之網路首先利用一前置處理(Minimum Spanning Tree [ ])擷取基因資料點之間垂直隸屬關係,在隨後叢集分析的過程中,該些隸屬關係結合區域統計特徵(如:Error-Tolerant Frequent Itemsets [22]、Silhouette width [23]等)以進行即時驗證目前(current)分群結果,本方法主要技術特徵在於同時進行Clustering 及Validation,達到大幅改進運算速度及叢集可信度之目的。
This project consist of two major parts, (1) applying clustering task to a vast number of multi-dimensional microarray data, and (2) developing noval neural networks that incorporates with Modified Minimum Spanning Tree (MMST). Note that the neural networks used in our project can apply to tasks other than the clustering pre-process.
It is important to make sure that the human incurred errors and undesired noises should be removed prior to actually analyzing microarray data. For this, we have developed a Scale Shrinking Higher-Order Filter, SSHF [1] which differs conventional methods in that SSHF operates on a digitalized microarray image, instead of the original microarray image. In addition, in order to deal with the increasingly huge amount of microarray data and the vast variations among the different dimensions, we propose a self-organizing fusion neural network, SOFNN[2], which is effective in reducing the variations between the multi-dimensional microarray data.
After filtering out the noise and reducing the input variations through the uses of SSHF and SOFNN, we employ MMST to conduct the final clustering process.
Relation: NSC94-2213-E019-014
URI: http://ntour.ntou.edu.tw/ir/handle/987654321/11203
Appears in Collections:[電機工程學系] 研究計畫

Files in This Item:

File Description SizeFormat
index.html0KbHTML63View/Open


All items in NTOUR are protected by copyright, with all rights reserved.

 


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback