[논문]BCI에서 기계 학습을 위한 간질 뇌파 특징 선택을 통한 차원 감소 방법 분석

양통; 임창균

doi:10.13067/jkiecs.2018.13.6.1333

초록
AI-Helper

지금까지 뇌파(Electroencephalography - EEG)는 뇌전증 진단 및 치료를 위한 가장 중요하고 편리한 방법이었다. 그러나 뇌전증 뇌파 신호의 파형 특성은 매우 약하고 비 정지 상태이며 배경 노이즈가 강하기 때문에 식별하기가 어렵다. 이 논문에서는 간질 뇌파의 특징 선택을 통한 차원 감소를 통한 분류 방법의 효과를 분석한다. 우리는 차원 감소를 위해 주 요소 분석, 커널 요소 분석, 선형 판별 분석 방법을 사용하였다. 차원 감소방법의 성능 분석을 위해 Support Vector Machine: SVM), Logistic Regression(: LR), K-Nearestneighbor(: K-NN), Decision Tree(: DR), Random Forest(: RF) 분류 방법들을 사용해 평가하였다. 실험 결과에 따르면, PCA는 SVM, LR 및 K-NN에서 75% 정확도를 나타냈다. KPCA는 SVM과 K-KNN에서 85%의 성능을 보였으며 LDA는 K-NN를 이용했을 때 100 %의 정확도 보여주었다. 따라서 LDA를 이용한 차원 감소가 뇌전증 EEG 신호에 대한 최고의 분류 결과 보여주었다.

Abstract ▼ AI-Helper

Until now, Electroencephalography(: EEG) has been the most important and convenient method for the diagnosis and treatment of epilepsy. However, it is difficult to identify the wave characteristics of an epileptic EEG signals because it is very weak, non-stationary and has strong background noise. I...

Until now, Electroencephalography(: EEG) has been the most important and convenient method for the diagnosis and treatment of epilepsy. However, it is difficult to identify the wave characteristics of an epileptic EEG signals because it is very weak, non-stationary and has strong background noise. In this paper, we analyse the effect of dimensionality reduction methods on Epileptic EEG feature selection and classification. Three dimensionality reduction methods: Pincipal Component Analysis(: PCA), Kernel Principal Component Analysis(: KPCA) and Linear Discriminant Analysis(: LDA） were investigated. The performance of each method was evaluated by using Support Vector Machine SVM, Logistic Regression(: LR), K-Nearestneighbor(: K-NN), Decision Tree(: DR) and Random Forest(: RF). From the experimental result, PCA recorded 75% of highest accuracy in SVM, LR and K-NN. KPCA recorded 85% of best performance in SVM and K-KNN while LDA achieved 100% accuracy in K-NN. Thus, LDA dimensionality reduction is found to provide the best classification result for epileptic EEG signal.

주제어

표/그림 (8)

그림 Fig. 1 Electroencephalograms in normal brain for two 3s continuous EEG Signals. It depicts the EEG of the brain under normal condition which is chaotic and irregular.
그림 Fig. 2 Electroencephalograms in patients with epilepsy for two 3s continuous EEG Signals. It depicts the EEG of the brain during Epilepsy attack. It has higher amplitude and rhymic pattern than under normal condition.
표 Table 1. The composition of training set sample points and test set sample points
그림 Fig. 3 System structure. The system consist of two major parts:- Dimensionality reduction methods and classifiers. The dimensionality reduction methods include PCA, KPCA and LDA while the classifiers used include SVM, LR, KNN, DT, and R
그림 Fig. 4 General flow chart. The features from the dataset are extracted using discrete wavelet transform then feature selection were conducted using three dimensionality reduction methods before passing the data to classifiers.
그림 Fig. 5 Comparison of the performance of dimensionality reduction methods and descret wave transformation in classification
그림 Fig. 6 Comparing epileptic EEG classification results with three dimension reduction methods
표 Table 2. Accuracy of dimensionality reduction methods with classifiers

AI 본문요약
AI-Helper

* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.

가설 설정

2) Most of dbN have no symmetry. For some wavelet functions, asymmetry is very obvious.
3) Regularity increases with the number N.
4) Function has orthogonality.

제안 방법

In this research, we have successful demonstrated that employing dimensionality reduction methods to EEG data can improve the performance of epilepsy classification. PCA, KPCA and LDA reduction methods were applied to the data after discrete wave transformation of the data.
In this work an epilepsy detection algorithm based on the PCA dimension reduction method and five different classifiers. First we determine the input N-dimensional sample D and the dimension N ′ to be reduced.
Its also refers to the method and process of using a computer to extract characteristic information in an image [6]. In this work, we analyze three feature selection methods for epileptic EEG recognition based on five classifiers and compare the results to identify the best method for Epilepsy classification.
4 General flow chart. The features from the dataset are extracted using discrete wavelet transform then feature selection were conducted using three dimensionality reduction methods before passing the data to classifiers.
Daubechies-2 wavelet transformation function was first applied to remove noise and extract 20 eigenvalues features. The transformed EEG was then subjected separately to three different dimensionality reduction methods to reduce the dimension of the EEG data. The reduction methods include Pincipal Component Analysis（PCA）, Kernel Principal Component Analysis (KPCA) and Linear Discriminant Analysis （LDA).
The reduction methods include Pincipal Component Analysis（PCA）, Kernel Principal Component Analysis (KPCA) and Linear Discriminant Analysis （LDA). Then each reduced data were then used for classification using Support Vector Machine (SVM), Logistic Regression (LR), K-Nearest Neighbor(KNN), Decision Tree (TR) and Random Forest(RF). Fig.
PCA, KPCA and LDA reduction methods were applied to the data after discrete wave transformation of the data. Then the reduced EEG data were each classified using Support Vector Machine (SVM, Logistic Regression (LR), K-Nearestneighbor (K-NN), Decision Tree (DR) and Random Forest (RF). From the experimental result, PCA recorded 75% highest accuracy in SVM, LR and K-NN.

대상 데이터

Each category has 100 samples. 80 samples were randomly selected from each class as training and 20 samples were used as test sets. Table 1 shows the composition of the dataset.
The data used in this study is a publicly available data from the Epilepsy Research Center at Bonn University in Germany [9]. The EEG data set consists of five subsets; denoted A~E, and each subset contains 100 segments of EEG signals. Each segment consists of 4097 sampling points.
The data used in this study is a publicly available data from the Epilepsy Research Center at Bonn University in Germany [9]. The EEG data set consists of five subsets; denoted A~E, and each subset contains 100 segments of EEG signals.

성능/효과

Then the reduced EEG data were each classified using Support Vector Machine (SVM, Logistic Regression (LR), K-Nearestneighbor (K-NN), Decision Tree (DR) and Random Forest (RF). From the experimental result, PCA recorded 75% highest accuracy in SVM, LR and K-NN. KPCA recorded 0.
From the experimental result, PCA recorded 75% highest accuracy in SVM, LR and K-NN. KPCA recorded 0.85% best performance in SVM and K-KNN while LDA achieved 100% accuracy in K-NN. Thus, LDA dimensionality reduction is found to provide the best classification result for epileptic EEG signal.

참고문헌 (13)

L. Rrambaiolli, N. Spolaor, and A. Corena, "Feature selection before EEG classification supports the diagnosis of Alzheimer's disease," Clinical Neurophysiology, vol. 128, no. 10, 2017, pp. 2058-2067.

상세보기
S. Moshe, E. Perucca, P. Ryvlin, and T. Tomson, "Epilepsy: new advances," The Lancet, vol. 385, Issue 9971, Mar. 2015, pp. 884-898.
J. Zhao, W. Zhou, K. Liu, and D. Cai, "EEG Signal Classification Method Based on Svm And Wavelet Analysis,"Computer Applications & Software, vol. 28, no. 5, 2011, pp. 114-116.
J. Jo "Drone Based Sensor Network Scenario for the Efficient Pedestrian's EEG Signal Transmission", J. of the Korea Institute of Electronic Communication Sciences, Sept. vol. 11, no. 9, 2016, pp. 923-928.
Y. Jang "Analysis of Concentration-Related EEG Component Due to Smartphone", J. of the Korea Institute of Electronic Communication Sciences, vol. 11, no. 7, July 2016, pp. 717-722.
G. Kumar and P. Bhatia, "A Detailed Review of Feature Extraction in Image Processing Systems," Fourth International Conference on Advanced Computing & Communication Technologies, Rohtak, India, Feb. 2014, pp. 5-12.
R. Davidson, D. Jackson, and C. Larson, Human electroencephalography, Handbook of Psychophysiology, New York. Cambridge University Press, 2000.
M. Nurujjaman, R. Narayanan, and A. Sekar Iyengar, "Comparative study of nonlinear properties of EEG signals of normal persons and epileptic patients," Nonlinear Biomedical Physics, vol. 3, no. 1, 2009, pp. 1-5.
R. Andrzejak, K. Lehnertz, C. Rieke, F. Mormann, P. David, and C. Elger, "Indications of nonlinear deterministic and finite dimensional structures in time series of brain electrical activity: Dependence on recording region and brain state," Physical Review E, vol. 64, 061907, Nov. 2001, pp. 1-8. view
Q. Cai , H. Chen, and L. Xie, "Analysis of EEG Based on Improvement Wavelet Transform," Computer Technology & Development, 2008.
J. Costa, P. Da-Silva, R. Almeida, and A. Infantosi, "Validation in Principal Components Analysis Applied to EEG Data," Computational and Mathematical Methods in Medicine, vol. 2014, Sep.2014, pp. 1-10.
L. Cao, K. Chua, K. Chongc, H. Lee, and Q. M. Gu, "A comparison of PCA, KPCA and ICA for dimensionality reduction in support vector machine," Neurocomputing, vol. 55, issues 1-2, Sept. 2003, pp. 321-336.
A. Subasi and M. I. Gursoyb, "EEG signal classification using PCA, ICA, LDA and support vector machines," Expert Systems with Applications, vol. 37, issue 12, Dec. 2010, pp. 8659-8666.

상세보기

이 논문을 인용한 문헌

저자의 다른 논문 :

LOADING...

내보내기 구분	파일저장 인쇄 메일전송
구성항목	기본정보 상세정보 관리번호, 논문명, 저널/프로시딩명, 저자 , 발행년, 권, 호, 시작페이지, 끝페이지, 발행기관 관리번호, 논문명, 대등논문명, 저자 , 저널/프로시딩명, 발행기관, 발행년, 발행언어, 권, 호, 시작페이지, 끝페이지, ISBN, ISSN, 주제분야, 키워드, 초록(한글), 초록(영문), 저자(소속기관)
저장형식	Text(ASCII format) Excel format RefWorks Direct Export RIS format (for Reference Manager, ProCite, EndNote), Scholar's Aids, Mendeley
메일정보	받는사람 (필수) @ 보내는사람 (선택) @ 제목 내용 KISTI 검색결과 이메일 서비스
안내	총 건의 자료가 검색되었습니다. 다운받으실 자료의 인덱스를 입력하세요. (1-10,000) 검색결과의 순서대로 최대 10,000건 까지 다운로드가 가능합니다. 데이타가 많을 경우 속도가 느려질 수 있습니다.(최대 2~3분 소요) 다운로드 파일은 UTF-8 형태로 저장됩니다. 파일의 내용이 제대로 보이지 않을실 때는 웹브라우저 상단의 보기 -> 인코딩 -> 자동선택 여부를 확인하십시오. ~ Text(ASCII format) Excel format

연합인증

BCI에서 기계 학습을 위한 간질 뇌파 특징 선택을 통한 차원 감소 방법 분석
Analysis of Dimensionality Reduction Methods Through Epileptic EEG Feature Selection for Machine Learning in BCI 원문보기

초록
AI-Helper

Abstract ▼ AI-Helper

주제어

표/그림 (8)

표/그림 (8)

AI 본문요약
AI-Helper

가설 설정

제안 방법

대상 데이터

성능/효과

참고문헌 (13)

이 논문을 인용한 문헌

저자의 다른 논문 :

연구과제 타임라인

관련 콘텐츠

원문 보기

원문 URL 링크

오픈액세스(OA) 유형

이 논문과 함께 이용한 콘텐츠

AI-Helper ※ AI-Helper는 오픈소스 모델을 사용합니다.

선택된 텍스트

연합인증

BCI에서 기계 학습을 위한 간질 뇌파 특징 선택을 통한 차원 감소 방법 분석 Analysis of Dimensionality Reduction Methods Through Epileptic EEG Feature Selection for Machine Learning in BCI 원문보기

초록 용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

주제어

표/그림 (8) 모든 표/그림 보기

표/그림 (8) 슬라이드로 보기

AI 본문요약 엑셀 다운로드 AI-Helper

가설 설정

제안 방법

대상 데이터

성능/효과

참고문헌 (13)

이 논문을 인용한 문헌

저자의 다른 논문 :

임창균 (13)

연구과제 타임라인

전체(0) 논문(0) 특허(0) 보고서(0)

전체(0) 논문(0) 특허(0) 보고서(0)

관련 콘텐츠

원문 보기

원문 URL 링크

오픈액세스(OA) 유형

이 논문과 함께 이용한 콘텐츠

AI-Helper ※ AI-Helper는 오픈소스 모델을 사용합니다.

선택된 텍스트

BCI에서 기계 학습을 위한 간질 뇌파 특징 선택을 통한 차원 감소 방법 분석
Analysis of Dimensionality Reduction Methods Through Epileptic EEG Feature Selection for Machine Learning in BCI 원문보기

초록
AI-Helper

표/그림 (8)

표/그림 (8)

AI 본문요약
AI-Helper