National Taiwan Ocean University Institutional Repository:Item 987654321/11253
English  |  正體中文  |  简体中文  |  Items with full text/Total items : 26999/38800
Visitors : 2398826      Online Users : 66
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/987654321/11253

Title: 結合神經網路與HHT於腦波信號之判讀
Incorporation of Neural Networks and HHT for Recognition of EEG Signals
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;希爾伯特-黃轉換;圖樣識別;分類器
Neural Network;Hilbert-Huang Transformation;Pattern Recognition;Classifier
Date: 2009-08
Issue Date: 2011-06-28T08:08:48Z
Publisher: 行政院國家科學委員會
Abstract: 摘要:腦波訊號之頻率關聯性或時間-空間特性係非穩態(non-stationary)、非線性,具高變異性,不易精確偵測EEG信號瞬時事件(transient event),本計畫擬使用希爾伯特-黃(HHT) [3-5] 轉換技術擷取腦波訊號之時頻特徵向量作為訓練神經網路(Neural Network)之輸入向量,配合神經網路的學習能力,將以三個不同spatio-temporal之神經網路模型,搭配前述HHT 時頻特徵資料,並以預先建立的驗證資料庫來進行訓練測試,各別網路訓練完畢後,再施以一微調訓練,即允許不同神經網路間互相學習,相較於傳統訓練單一神經網路,可更有效增加神經網路之概泛能力進而降低EEG高變異性的影響以獲得最適網路架構(層數及各層神經元數目)。預期由於神經網路對於雜訊有一定的強建性,因此依據輸入多通道腦波信號,學習後的神經網路可作為一個強建性分類器(classifier)或腦波異常量化器。採用神經網路於偵測並判別異常波訊號,其特色係不須假設特徵向量機率密度函數之型式為已知,藉由訓練神經網路學習偵測辨識腦波訊號,輔助癲癇病症診斷,提升醫療品質。
abstract:This research project aims to develop a technique useful in detecting multi-channel EEG transient events or characteristic waves, such as spikes, spindle, hump, etc. Our approach is based on utilizing a set of Hilbert-Huang Transformation (HHT) feature data to train three different spatio-temporal neural networks. Key essence of the training process is it allows inter-learning among the three neural networks in order to further enhance the generalization capability of the three neural networks for building a characteristic waves classifier or transient event detector. The goal of training is set out to be threefold: (1) to exploit the structural relation between EEG characteristic waves and the HHT time-frequency feature data by utilizing capability of spatio-temporal neural networks in exploiting the non-stationary and non-linear signals (2) to determine near-optimal network architecture for EEG transient event detection and classification (3) finally but most importantly, to reduce the effect of the large variation nature of EEG signals in detecting and classifying transient events by using the generalization capability of neural networks.
Relation: NSC98-2221-E019-032
URI: http://ntour.ntou.edu.tw/ir/handle/987654321/11253
Appears in Collections:[Department of Electrical Engineering] Research Reports

Files in This Item:

File Description SizeFormat
index.html0KbHTML100View/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