• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 추계학술대회
• /
• pp.207-209
• /
• 2022

IoT환경에서 스마트 디바이스로부터 사람의 신체 활동을 인식하여 생활 패턴 데이터를 수집할 수 있게 되었다. 본 논문에서는 제안된 모델은 예측단계와 추천단계로 구성한다. 예측 단계는 생활 패턴 데이터로부터 수집된 데이터셋을 기계학습을 통해 로지스틱 회귀와 k-최근접 이웃 알고리즘을 활용하여 불안과 우울의 척도를 예측한다. 추천 단계는 불안과 우울 증상으로 분류된 경우 이를 호전시킬 수 있는 음식과 가벼운 운동을 추천하기 위해 주성분 분석 알고리즘을 적용한다. 제안한 불안·우울 예측과 음식·운동 추천은 개인의 삶의 품질 개선에 파급효과가 있을 것으로 기대한다.

• Communications for Statistical Applications and Methods
• /
• 제31권2호
• /
• pp.213-234
• /
• 2024

In recent decades, increasing research attention has been directed toward predicting the price of stocks in financial markets using deep learning methods. For instance, recurrent neural network (RNN) is known to be competitive for datasets with time-series data. Long short term memory (LSTM) further improves RNN by providing an alternative approach to the gradient loss problem. LSTM has its own advantage in predictive accuracy by retaining memory for a longer time. In this paper, we combine both supervised and unsupervised dimension reduction methods with LSTM to enhance the forecasting performance and refer to this as a dimension reduction based LSTM (DR-LSTM) approach. For a supervised dimension reduction method, we use methods such as sliced inverse regression (SIR), sparse SIR, and kernel SIR. Furthermore, principal component analysis (PCA), sparse PCA, and kernel PCA are used as unsupervised dimension reduction methods. Using datasets of real stock market index (S&P 500, STOXX Europe 600, and KOSPI), we present a comparative study on predictive accuracy between six DR-LSTM methods and time series modeling.

• 대한교통학회지
• /
• 제26권3호
• /
• pp.53-66
• /
• 2008

본 연구는 기존의 도로기능 분류체계를 보완하면서 유형화된 도로구간들의 교통특성을 규명하기 위해 새로운 관점에서 도로특성분류 개념을 정립하였다. 도로특성분류는 교통계획, 교통운영관리 등의 교통전반으로 설계 및 정책을 수립하고 지침을 마련하는데 중요한 판단자료로 이용될 것으로 기대된다. 또한, 도로특성분류를 위해 일반국도 상시지점 조사자료를 토대로 12개의 설명변수를 산출하였으며, 이 설명변수들간의 상호상관을 통한 잠재구조 및 다중공선성 검토, 요인점수를 추출하는 탐색적 요인분석을 수행하였다. 연구 방향은 탐색적 요인분석의 각 실행단계별 접하게 되는 의사결정 문제를 세밀하게 검토하였으며, 각 논점별로 올바른 평가기준 방법을 제시하여 최종적인 종합결론을 도출하였다. 적정 설명변수와 요인 수를 결정하기 위해 10개의 시나리오를 비교분석한 결과, 처음 제시한 12개의 설명변수를 모두 포함한 경우가 가장 우수한 것으로 분석되었으며, 4개의 요인이 가장 적정한 것으로 나타났다. 본 연구결과는 추후에 다양한 분석방법(군집분석, 회귀분석, 판별분석 등)에 있어서 객관적인 입력자료로 사용됨에 따라 보다 정확한 연구결과가 도출될 것으로 기대된다.

• Journal of Biosystems Engineering
• /
• 제34권3호
• /
• pp.197-203
• /
• 2009

This study was performed to develop a nondestructive determination technology for sugar content in sweet persimmons, and the main research results included the following. In order to determine sugar content in sweet persimmons, a dual side reflex was adopted, and the study was to measure sugar content using a reflectance spectrum for 2 parts because it was difficult to determine representative sugar content due to a great deviation in sugar content according to the part of sweet persimmons. To predict sugar contents of sweet persimmon, PLSR and PCR models were compared with a few preprocess methods. As a result, PLSR had $R^2$=0.67, SEP=0.42 brix, LV=11, and PCR had $R^2$=0.65, SEP=0.41 brix, PC=16. SNV method was the best among preprocess methods for predicting sugar contents.

• 한국조경학회지
• /
• 제22권3호
• /
• pp.151-162
• /
• 1994

This is a comparative study on the use pattern and satisfaction of skiers between the suburban skiing ground and the resort one. The purpose of this study is to provide basic data for the planning and the management of skiing ground. The sites of case study are Yong Pyung Ski Resort (Resort type) and Bears Town(Suburban type). Data were collected from questionaire. A total of 420 questionaires were completed. And data were subjected to following analysis: First, the descriptive statistics(mean, chi-square analysis etc.) were used to compare the characteristics of the users and the use pattern of two sites. Second, factor analysis was utilized to reduce 22 satisfaction items into the smaller number of factors. Third, regression analysis was used to find the factors affecting users' overall satisfaction in each skiing ground. The findings of this study are as follows: First, it was proved that the characteristics of users between tow sites were different in terms of age, income, and skill level. Second, it was proved that the use pattern between two sites were different in terms of travel distance from home, traffic mode, length of stay, accommodation type, and the money spent per day. Third, By a principal component factor analysis several factors of satisfaction are found: In physical terms, they are 'slope and life facilities', 'recreation and lodge facilities', 'accessibility', 'crowding', and 'landscape'. In psychological terms, they are 'skiing skills and thrills', and 'relaxation and freedom'. Forth, As the result of the stepwise regression analysis, it was yielded that 'relaxation/and freedom' was most important factor to predict the overall satisfaction in both skiing ground. And it was proved that not only physical factors but also phychological(need gratifying) factors were important sources of the satisfaction.

• Journal of Biosystems Engineering
• /
• 제41권3호
• /
• pp.208-220
• /
• 2016

Purpose: Fusel oil is a potent volatile aroma compound found in many alcoholic beverages. At low concentrations, it makes an essential contribution to the flavor and aroma of fermented alcoholic beverages, while at high concentrations, it induced an off-flavor and is thought to cause undesirable side effects. In this work, we introduce Fourier transform near-infrared (FT-NIR) spectroscopy as a rapid and nondestructive technique for the quantitative determination of fusel oil in the Korean alcoholic beverage "soju". Methods: FT-NIR transmittance spectra in the 1000-2500 nm region were collected for 120 soju samples with fusel oil concentrations ranging from 0 to 1400 ppm. The calibration and validation data sets were designed using data from 75 and 45 samples, respectively. The net analyte signal (NAS) was used as a preprocessing method before the application of the partial least-square regression (PLSR) and principal component regression (PCR) methods for predicting fusel oil concentration. A novel variable selection method was adopted to determine the most informative spectral variables to minimize the effect of nonmodeled interferences. Finally, the efficiency of the developed technique was evaluated with two different validation sets. Results: The results revealed that the NAS-PLSR model with selected variables ($R^2_{\upsilon}=0.95$, RMSEV = 100ppm) did not outperform the NAS-PCR model (($R^2_{\upsilon}=0.97$, RMSEV = 7 8.9ppm). In addition, the NAS-PCR shows a better recovery for validation set 2 and a lower relative error for validation set 3 than the NAS-PLSR model. Conclusion: The experimental results indicate that the proposed technique could be an alternative to conventional methods for the quantitative determination of fusel oil in alcoholic beverages and has the potential for use in in-line process control.

• 한국연초학회지
• /
• 제17권2호
• /
• pp.177-183
• /
• 1995

This study was conducted to develop the most precise NIR(near infrared spectrometric) calibration for rapid determination of chemical composition in ground samples of toasted burley tobacco using stepwise, stepup, principal component regression(PCR), partial least square(PLS) and modified partial least square(MPLS) calibration method. The number of wavelength(W) selected by stepup multiple linear regression using: second derivative spectra was as follows: total sugar(TS)-4 W, nicotine-9 W, total nitrogen(TN)-2 W, ash-8 W, total volatile base(TVB)-5 W, chlorine4 W, L of color-6 W, a of color-6 W and b of color-7 W. Comparing the calibration equations followed by each chemical components, the most precise calibration equation was MPLS for 75, a and b of color, PLS for nicotine, ash, TVB, chlorine and L of color and stepup for TN. The standard error of calibration(SEC) and standard error of performance(SEP) between result of near infrared analysis and standard laboratory analysis were 0.18, 0.40% for 75, 0.06, 0.08% for nicotine, 0.18, 0.16% for TN, 0.33, 0.46% for ash, 0.04, 0.03% for TVB, 0.08, 0.06% for chlorine, 0.54, 0.58 for L of color, 0.22, 0.22 for a of color and 0.27, 0.27 for b of color, respectively. The SEC and SEP of ash and TVB were within allowable error of standard laboratory analysis, nicotine, TN and chlorine were 1.2-2.0 times and 75 were 2.1-4.0 times larger than allowable error of standard laboratory analysis. The ratio of SEC and SEP to mean were 1.5, 1.6% for L of color, 3.7, 3.8% for a of color and 1.8, 1.8% for b of color, respectively. Key words : burley tobacco chemistry, near infrared spectroscopy.

• 동아시아식생활학회지
• /
• 제22권6호
• /
• pp.836-850
• /
• 2012

This study was conducted to identify the sensory characteristics of the Korean traditional confectionery, dasik, prepared under different conditions and to compare their consumer acceptance in Korea. To accomplish this, descriptive analysis of eight samples prepared using two types of rice cake powder, dasik (Rflour, Rflour_Omija), brown rice powder red ginseng dasik (Brice_Ginseng_P), pinepollen dasik (PineP), black sesame dasik (BSesame), bean dasik (Rbean), and two types of mungbean starch dasik (Starch_Omija, Starch_Greentea), was conducted by ten trained panelists. In addition, 81 consumers evaluated the overall acceptance (OL), acceptance of appearance (APPL), odor (ODL), flavor (FLL), and texture (TXTL) of the samples using a 9-point hedonic scale, as well as the perceived intensities of sesame flavor, sweetness, and hardness using a 9-point just-about-right (JAR) scale. Partial least square- regression (PLSR) indicated that the BSesame and Rbean samples, which had significantly (p<0.05) high roasted sesame, burnt, greasy, glossy, and cooked chestnut flavor scores, had the highest acceptability and consumer desire scores. Additionally, the PineP and Rflour_Omija samples, which had relatively high particle size, transparency, roughness, spoiled tofu, fermentation and raw rice flavor scores, were the least preferred samples. Therefore, roasted sesame, burnt, greasy, glossy, and cooked chestnut flavor attributes were considered drivers of "liking" whereas particle size, transparent, roughness, spoiled tofu, fermentation, and raw rice flavor attributes acted as drivers of "disliking" among consumers.

• 전기전자학회논문지
• /
• 제19권4호
• /
• pp.541-547
• /
• 2015

현재 반도체 제조 공정에서 집적회로의 소자 크기가 점점 작아짐에 따라 플라즈마 식각 공정에서의 식각 종료점 검출이 더 어려워지고 있다. 식각 종료점 검출은 위해서는 반도체 장비에 다양한 종류의 센서를 설치하고 이 센서를 통해 데이터를 얻고 분석해야 한다. 기존의 식각 종료점 검출 방식은 주로 By-product의 OES 데이터를 분석하여 진행되었는데 본 연구에서는 By-product 와 Etchant gas 의 OES 데이터를 함께 분석하여 식각 종료점 검출 결과에 신뢰성을 더 높이고자 하였다. 또한, 데이터 분석을 위해 OES-SNR, PCA, Polynomial Regression, eHMM 등의 기법들을 사용하여 진행하였다.

• 방송공학회논문지
• /
• 제24권2호
• /
• pp.264-272
• /
• 2019

본 논문은 비디오로부터 추출한 프레임으로부터 3차원 인체 형상과 자세 복원을 수행하고 이를 시간 축에서 자연스럽고 부드러운 움직임을 나타내도록 보정하는 기법을 제안한다. 제안하는 기법은 우선 비디오로부터 추출한 개별 프레임으로부터 convolutional neural network을 이용하여 관절의 위치와 인체의 윤곽을 추정한다. 인체의 형상 및 자세는 매개변수 기반의 3차원 변형가능 모델(morphable model)을 2차원 영상으로 투영후 정합하여 최적의 매개변수 값을 추정한다. 이 때 각 프레임에 대한 복원이 개별적으로 수행되면 시간 축에서 자세의 연속성과 체형의 일관성이 보장되지 못하고 올바르지 못한 복원 결과가 나타난다. 제안하는 기법은 이러한 문제점을 보완하기 위하여 각 프레임으로부터 복원된 3차원 변형가능 모델의 주성분 매개변수의 분석 및 보간을 수행한다. 실험결과 3차원 인체 복원에 오류가 발생한 프레임에 대해 이전과 이후 프레임들 사이의 관계를 통해 오류가 보정되어 개선된 복원 결과를 얻을 수 있음을 보인다.