동공 크기 변화 기반 효과적 학습 상태 분류를 위한 데이터 재구성 알고리즘
Data Reconfiguration Algorithm for Efficient Learning State Classifications Based on Pupil Sizes
- 주제(키워드) Machine Learning , Data Integration , Classification Model , Support Vector Machine , K-nearest neighbor , Time Series Data , 기계학습 , 데이터 통합 , 분류모델 , 서포트 벡터 머신 , K최근접이웃 , 시계열 데이터
- 주제(기타) 전자/정보통신공학
- 설명문(URI) https://www.kci.go.kr/kciportal/ci/sereArticleSearch/ciSereArtiView.kci?sereArticleSearchBean.artiId=ART002637878
- 등재 KCI등재
- 발행기관 한국통신학회
- 발행년도 2020
- URI http://www.dcollection.net/handler/ewha/000000176259
- 본문언어 한국어
- Published As http://dx.doi.org/10.7840/kics.2020.45.10.1756
초록/요약
기계학습, 데이터 통합, 분류모델, 서포트 벡터 머신, K최근접이웃, 시계열 데With the recent development of smart devices, the use of wearable devices became popular, which enabled us to collect individual biometric data such as heart rate, energy consumption and number of steps smoothly. As high performance and high accurate sensors are equipped in wearable devices, the more biometric data can be collected. Since the biometric data are less susceptible to racial and cultural differences, they can be used as a reliable set of data sources. Because of inevitable noises included in the processes of data acquisition, however, the performance of the machine learning algorithms employed for efficiently processing the biometric data and accurately analyzing them can be limited. In this paper, we focus on the sizes of pupil data among a variety of biometric data, which are used as an indicator of concentration. We adopt machine learning based classification algorithms for inferring the degree of concentration by analyzing the time-varying changes of pupil sizes. In order to improve the classification performance, we propose a pre-processing algorithm for data reconfiguration. Our simulation and experiment results confirm that the proposed pre-processing for data reconfiguration improves the data classification performance.이터
more