English  |  正體中文  |  简体中文  |  Items with full text/Total items : 26988/38789
Visitors : 2345085      Online Users : 37
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/987654321/49303

Title: 利用GPU加速模糊k均值分群演算法
Speeding up Fuzzy k-Means Clustering Algorithm on GPU
Authors: Chang, Tsun-Chuan
張宗荃
Contributors: NTOU:Department of Computer Science and Engineering
國立臺灣海洋大學:資訊工程學系
Keywords: 圖形處理器;資料分群;模糊k均值分群法
GPU;CUDA;data clustering;Fuzzy k-Means
Date: 2015
Issue Date: 2018-08-22T06:56:27Z
Abstract: 近幾年由於資料量的增加,在計算效率需求增加之下硬體也逐漸地發展,圖形處理器(GPU)即為其中之一。現階段,GPU擁有大規模的平行處理架構,因此能夠具備比中央處理器(CPU)更快速的計算浮點數能力,進而成為強而有力的設備供使用者去運用。資料分群在資料探勘中也是一個重要的議題,應用的範圍也相當廣泛,在本篇論文中,我們將使用GPU實作一個資料分群中最常見的演算法-模糊k均值分群法。在計算membership時我們使用了兩種不同的策略以確保所有的測試範例都能在最佳化的狀況下執行。在計算新的群中心時,我們也因應不同的群中心大小而使用兩種方法。在本篇論文中,我們將會比較模糊k均值分群法在CPU和GPU運算後的速度,同時也會分析使用GPU運算此分群法的利與弊。從我們的實驗結果可以得知我們提出的方法可以大量減少模糊k均值分群法的計算時間,約為六到十三倍。
Recently the hardware aspect of the computer has gradually become more developed and advanced in order to keep up with the rise in data and computational efficiency. The graphics processing unit (GPU) is one of the hardware that has followed that trait. The ability of the GPU to do floating point operations has exceeded that of the CPU due to the massive parallel architecture within the GPU, which has also become a powerful technology for programming purposes. Data clustering is also an important issue in data mining, but the applications of data clustering are also very wide. In this thesis we will use the GPU to implement one of the most common algorithms of data clustering, known as fuzzy k-means clustering (FKM). In FKM Clustering, when updating memberships, we provide two different strategies to ensure that all test samples can be executed optimally, and when calculating new centroids, we use two different ways which corresponds to the different centroids sizes. We will also make comparisons with CPU based implementations and analyse the pros and cons of using GPU. Our experimental results show that our GPU-based FKM algorithms are about six to thirteen times faster than the optimized CPU code-based FKM algorithms.
URI: http://ethesys.lib.ntou.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=G0010257038.id
http://ntour.ntou.edu.tw:8080/ir/handle/987654321/49303
Appears in Collections:[資訊工程學系] 博碩士論文

Files in This Item:

File Description SizeFormat
index.html0KbHTML3View/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