Earticle

본 연구의 목적은 여행사가 서비스교육을 위해 제공하는 투입변인(서비스교육 프로그램, 교육지원)이 과정변인(목표, 교수자, 학습자, 내 용․환경)에 어떠한 영향을 미치는지, 또한 이러한 과정변인들이 성과변인(비재무성과, 고객성과)에 어떠한 영향을 미치는지를 알아보고자 하였다. 본 연구는 서울․부산지역의 여행사에 근무하고 있는 직원들을 대상으로 설문조사를 실시하였으며 260부를 배부하여 회수된 214부를 SPSS 25.0 프로그램을 활용하여 다중회귀분석을 통해 변인 간의 인과관계를 규명하고자 하였다. 분석결과를 보면 교육지원과 서비스교육 프로그램은 과정변인(목표, 교수자, 학습자, 내용․환경)에 통계적으로 유의하게 나타났다. 학습자, 내용․환경변인은 비재무성 과에 영향을 미치는 것으로 나타났고 내용․환경, 교수자변인은 고객성과에 통계적으로 유의한 것으로 나타났다. 직원들은 그들의 조직이 제공하는 서비스교육이 중요하다고 인지하고 있으며 지속적인 서비스교육은 조직에 대한 직원의 만족감이 향상되고 직원이직률을 감소시 키는 중요한 요소임을 확인하였다. 또한, 서비스교육을 통해 신규고객을 확보할 수 있는 근거로서 여행사 직원과 고객과의 성공적인 접촉점 을 형성하는 데 도움이 될 것이라고 사료된다.

The purpose of this study was to investigate the relationship of the input variable, the influence on process variables, and the influence those variables have on performance variable within the travel agency. In all, 214 questionnaires collected from the employees who were working at travel agencies were returned out of 260 questionnaires distributed in Seoul and Busan. This study implemented multiple regression analysis to examine how employees perceive the service education provided by the organization utilizing the SPSS 25.0 program. The results of this study demonstrated that education support and service education programs were statistically significant in process variables(objective, teachers, learners, content․environment). It was found that learners, content․environment variables had an effect on non-financial performance, and content․environment and teachers variables were statistically significant in customer performance variables. Through this study, it was confirmed that continuous service education is an important factor in improving employee satisfaction and reducing employee turnover, and it is believed that it will help to form a successful contact point between travel agency employees and customers as the basis for attracting new customers in the field.

본 논문은 입력 전력 신호와 결정지향 오차 신호에 따라 스텝 크기가 변하는 가변 스텝 크기를 가지는 MSAGF-MMA의 성능 분석에 관한 것이다. 제안한 알고리즘은 변형된 Stop-and-Go 알고리즘으로 부터 얻어진 이진 플래그를 가지는 MMA에 스텝 크기가 입력 신호의 변동에 영향을 덜 받도록 함으로써 정상상태로의 수렴 속도를 안정하 게 높일 수 있는 입력 전력 신호에 따라 변하도록 하였다. 그와 동시에 정상상태에서 잔류 오차가 작아질 수 있도록 결정지향 오차 신호에 따라 스텝 크기가 가변될 수 있도록 하였다. 컴퓨터 모의실험 결과, 제안한 알고리즘이 MMA 및 MSAGF-MMA와 비교하여 정상상태로의 수렴 속도 면에서 뿐만 아니라 정상상태에서의 잔류 ISI와 averaged-MSE 등의 평가에서 매우 뛰어난 성능을 가짐을 확인하였다.

This paper is concerned with the performance analysis of MSAGF-MMA with variable step size whose step size varies according to input power signal and decision-directed error signal. The proposed algorithm is made to change according to the input power signal which can reliably increase the convergence speed to the steady state by making the step size less affected by the fluctuation of the input signal in the MMA having the binary flag obtained from the modified Stop-and-Go algorithm. At the same time, the step size can be varied according to the decision-directed error signal so that the residual error can be reduced in the steady state. As a result of computer simulations, it is confirmed that the proposed algorithm has a very good performance in the evaluation of residual ISI and averaged-MSE in steady state as well as in terms of convergence speed to steady state compared to MMA and MSAGF-MMA.

본 연구의 목적은 다음과 같이 요약될 수 있다. (1) 유발계수행렬에서 투입유발액과 생산유발액의 차이가 발생하는 근본적인 원인을 다양한 방법으로 규명하고자 한다. (2) 특히 최종수요(f)를 원인변수로 한 투입유발액과 최종수요를 원인변수로 한 생산 및 최종산출유발액 사이의 인과관계를 밝히고, 그 차별적인 성격을 비교 분석하는 것이다. (3) 유도된 변수 사이의 인과관계식을 토대로 하여, 투입유발액과 생산 및 최종산출유발액에 대한 경험적 사례 연구를 수행함으로써 이들 유발액 사이의 실제적 차이를 규명하는 것이다. 실증 분석을 위한 기초 자료는『2013년 산업연관표』이며, 최종수요에 대한 생산유발계수행렬(Cf)은 국산과 수입을 통합한 Cf=(I-A)-1형으로 하였다. 유발계수행렬은 투입유발계수행렬과 생산유발계수행렬로 크게 대별된다. 투입유발계수행렬은 다시 산출물에 대한 투입유발계수행렬(Γ9)과 최종수요에 대한 투입유발계수행렬(Γf)로 구분된다. 생산유발계수행렬은 산출물에 대한 생산유발계수행렬(C9)과 최종수요에 대한 생산유발계수행렬(Cf)로 나누어진다. 원인변수가 최종수요(f)일 때에 최종수요에 대한 투입유발액(p)과 원인변수가 총산출(x)일 때에 산출물에 대한 투입유발액(q)을 각각 유도할 수 있다. 또한 원인변수가 최종수요(f)일 때에 생산유발액(x)과 최종산출유발액(o)을 각각 추계할 수 있다. 주요 분석 결과는 두 투입유발계수행렬과 두 생산유발계수행렬 사이의 피어슨 상관계수(γp)와 스피어맨 순위상관계수(γs)를 각각 구하면 모든 유발계수행렬 상호간에는 매우 높은 양의 상관관계를 보여주고 있다. 투입유발계수와 생산유발계수에서 제9부문(제1차 금속제품)이 유발효과 면에서 모두 제1순위를 차지하고 있다.

The specific objectives of this paper can be briefly summarized as follows. (1) We first clarify the differences in characteristics between input requirements and output requirements occur in the requirements matrices. (2) Based on a causal relation considering the final demand (f) as a cause variable, we conduct a comparative analysis between two input requirements (p and q ) for final demand (f) and gross output (x), and between (gross) output requirements (x) and final output requirements (o). (3) Through the formulas derived from the cause and effect variables, we empirically calculate input requirements, output requirements, and final output requirements from the 2013 national input-output (IO) tables compiled by the BOK and compare with the real differences among three different estimated variables of requirements. We then introduce the four requirements matrices: the two input requirements matrices Γf and Γ9 for final demand and (gross) output and the two output requirements matrices Cf and C9 for final demand and (gross) output, respectively by the decomposition by factors in the requirements matrices.

Statisticians, or data miners, are often requested to assess the importances of input variables in the given supervised learning model. For the purpose, one may rely on separate ad hoc measures depending on modeling types, such as linear regressions, the neural networks or trees. Consequently, the conceptual consistency in input variable importance measures is lacking, so that the measures cannot be directly used in comparing different types of models, which is often done in data mining processes, In this short communication, we propose a unified approach to the importance measurement of input variables. Our method uses sensitivity analysis which begins by perturbing the values of input variables and monitors the output change. Research scope is limited to the models for continuous output, although it is not difficult to extend the method to supervised learning models for categorical outcomes.

This study was carried out to evaluate the effect of feeding higher protein feeds with lesser amount, but feeding the constant total protein input for all treatments, on water quality and nitrite toxicity in channel catfish ponds. There was no significant difference in survival rate among treatments $(P>0.05)$. Specific growth rate (SGR) for Treatment 1$(28\%\;protein\;and\;100\%\;of\;satiation)$ was significantly higher $(P>0.05)$ than for Treatment 3$(36\%\;protein\;and\;87.5\%\;of\;satiation)$, but not significantly higher than for Treatment 2 $(32\%\;protein\;and\;77.8\%\;of\;satiation)$ at constant digestible energy (DE), 3.08kcal/g (treatments 1, 2 and 3). At constant DE/P (treatments 4, 2 and 5), no significant difference in SGR was observed among treatments. Feed conversion ratio (FCR) slightly improved or improved as dietary protein level increased from $28\%$ to $32\%$ and feed allowance decreased by $12.5\%$, but did not improve as dietary protein level increased from $32\%$ to $36\%$ and feed allowance decreased by $22.2\%$, at constant DE and constant DE/P. There was no significant difference in water quality variables, such as total ammonia nitrogen (TAN), nitrite, chlorophyll a, soluble phosphorous concentrations among treatments, but significant difference in water quality variables over time as amount of feed fed increased $(P<0.0001)$. There was a trend toward increase in TAN and nitrite over time. A strong linear regression was observed between mean total ammonia nitrogen and nitrite for all treatments Y (Nitrite) =$0.04\times (TAN)+0.01$, $R_2=0.89$. Methemoglobin percent in the blood of catifish was not significantly different among treatments. And its mean value was $7.5\%$, which was relatively low, so that it was not serious problem in catfish production pond under these experiment conditions. There was the stronger linear regression between the percentage of Methemoglobin and the molar ratio of nitrite to chloride rather than nitrite alone: $Y\;(Methemoglobin\;\%)\;=\;58.45\;\times\;(NO^{2-}/Cl^-)\;+\;0.41,\;R^2=0.60$. These results indicate that deterioration of water quality has no strong impact on poor weight gain for $36\%$ dietary protein in this study.

A UHF CMOS variable gain low-noise amplifier (LNA) is designed for mobile digital TV tuners. The proposed LNA adopts a feedback topology to cover a wide frequency range from 474 to 868 MHz, and it supports the notch filter function for the interoperability with the GSM terminal. In order to handle harmonic distortion by strong interferers, the gain of the proposed LNA is step-controlled while keeping almost the same input impedance. The proposed LNA is implemented in a $0.11{\mu}m$ CMOS process and consumes 6 mA at a 1.5 V supply voltage. In the measurement, it shows the power gain of greater than 16 dB, NF of less than 1.7 dB, and IIP3 of greater than -1.7 dBm for the UHF band.

분기 히스토리는 분기 예측기의 주된 입력 요소로 사용된다. 따라서 적절한 분기 히스토리의 사용은 분기 예측의 정확도 향상에 큰 영향을 미친다. 본 논문에서는 분기 예측의 정확도를 향상시키기 위한 방법의 하나로, 명령어의 연관성 분석을 통한 선별적 분기 히스토리 사용 기법을 제안한다. 우선, 본 논문에서는 명령어의 연관성을 분석하는 세 가지 서로 다른 알고리즘을 제안한다. 제안된 기법은 명령어의 레지스터 쓰기 연산에 기반하는 방법, 분기 명령어의 참조 레지스터에 기반하는 방법, 그리고 이들 두 가지 방식을 상호 결합하는 방법이다. 또한, 제안된 세 가지 알고리즘의 실질적 구현을 위해 이를 적용할 수 있는 가변 입력 gshare 예측기를 제안한다. 본 논문에서는 모의실험을 통해 세 가지 알고리즘의 특징 및 장단점을 비교 분석한다. 특히, 기존의 고정된 입력을 사용하는 방식과 비교하여 제안된 기법의 성능 향상의 정도를 분석하며, 사전 프로파일링을 통해 얻어진 최적의 입력에 대한 성능상의 차이도 소개한다.

Branch history is one of major input vectors in branch prediction. Therefore, the Proper use of branch history plays a critical role of improving branch prediction accuracy. To improve branch prediction accuracy, this paper proposes a new branch history management policy, based on interrelationship analysis of instructions. First of all, we propose three different algorithms to analyze the relationship: register-writhing method, branch-reading method, and merged method. Then we additionally propose variable input gshare predictor as an implementation of these algorithms. In simulation part, we provide performance differences among the algorithms and analyze their characteristics. In addition, we compare branch prediction accuracy between our proposals and conventional fixed input predictors. The performance comparison for optimal input branch predictor is also provided.

본 논문에서는 고정점 알고리즘의 독립성분분석과 상호정보 추정을 조합한 입력변수선택 기법을 제안하였다. 여기서 뉴우턴법에 기반을 둔 빠른 분석성능을 가지는 고정점 알고리즘의 독립성분분석은 입력변수 간의 독립성을 빠르게 찾기 위함이고, 입력변수의 확률밀도함수의 계산을 위해 적응적 분할을 이용한 상호정보의 추정은 변수상호간 종속성을 좀 더 정확하게 정량화하기 위함이다. 제안된 기법을 인위적으로 제시된 각 500개의 샘플을 가지는 6개의 독립신호와 1개의 종속신호를 대상으로 실험한 결과 빠르고 정확한 변수의 선택이 이루어짐을 확인하였다.

본 논문에서는 주요성분분석과 상호정보 추정을 조합한 입력변수선택 기법을 제안하였다. 여기서 주요성분분석은 2차원 통계성을 이용하여 입력변수 간의 독립성을 찾기 위함이고, 상호정보의 추정은 적응적 분할을 이용하여 입력변수의 확률밀도함수를 계산함으로써 변수상호간의 종속성을 좀더 정확하게 측정하기 위함이다. 제안된 기법을 인위적으로 제시된 각 500개의 샘플을 가지는 6개의 독립신호와 1개의 종속신호를 대상으로 실험한 결과, 빠르고 정확한 변수의 선택이 이루어짐을 확인하였다.

In this paper a wind power simulator is designed and implemented. This simulator realize the torque of wind blade by DC motor. And squirrel-cage induction machine is used as generator and controlled to achieve the maximum power point tracking(MPPT) algorithm.

Algal bloom is an ongoing issue in the management of freshwater systems for drinking water supply, and the chlorophyll-a concentration is commonly used to represent the status of algal bloom. Thus, the prediction of chlorophyll-a concentration is essential for the proper management of water quality. However, the chlorophyll-a concentration is affected by various water quality and environmental factors, so the prediction of its concentration is not an easy task. In recent years, many advanced machine learning algorithms have increasingly been used for the development of surrogate models to prediction the chlorophyll-a concentration in freshwater systems such as rivers or reservoirs. This study used a light gradient boosting machine(LightGBM), a gradient boosting decision tree algorithm, to develop an ensemble machine learning model to predict chlorophyll-a concentration. The field water quality data observed at Daecheong Lake, obtained from the real-time water information system in Korea, were used for the development of the model. The data include temperature, pH, electric conductivity, dissolved oxygen, total organic carbon, total nitrogen, total phosphorus, and chlorophyll-a. First, a LightGBM model was developed to predict the chlorophyll-a concentration by using the other seven items as independent input variables. Second, the time-lagged values of all the input variables were added as input variables to understand the effect of time lag of input variables on model performance. The time lag (i) ranges from 1 to 50 days. The model performance was evaluated using three indices, root mean squared error-observation standard deviation ration (RSR), Nash-Sutcliffe coefficient of efficiency (NSE) and mean absolute error (MAE). The model showed the best performance by adding a dataset with a one-day time lag (i=1) where RSR, NSE, and MAE were 0.359, 0.871 and 1.510, respectively. The improvement of model performance was observed when a dataset with a time lag up of about 15 days (i=15) was added.

신뢰성 분석은 불확실성으로 인한 제품의 성능 변동을 안전확률이나 파괴확률로 정량화 하여 설계에 이용하기 위해 연구되어 왔다. 불확실성은, 데이터의 양에 따라-물질의 본질적인 특성으로서의 많은 데이터가 주어진 경우의 물리적 불확실성과 부족한 데이터에서의 인식론적 불확실성으로 구분되고, 불확실성을 갖는 대상에 따라-입력변수 및 근사모델 불확실성으로 구분된다. 물리적 불확실성에 대한 연구는 많이 진행되어 왔지만, 실제 산업현장에는 부족한 데이터로 인한 인식론적 불확실성이 지배적이며 이에 대한 연구는 최근에서야 진행되고 있다. 불확실성을 고려하는 신뢰성 기반 설계에는 효율성을 위해 실제모델을 대체하는 근사모델이 이용되는데, 근사모델법 자체에 대한 연구는 많이 진행되어 왔으나, 근사모델 이기 때문에 존재하는 불확실성을 고려한 연구는 최근에서야 연구되기 시작하였다. 본 연구에서는 베이지안 접근법에 기반하여 입력변수 및 근사모델 불확실성을 통합 고려하는 새로운 신뢰성 분석 기법을 제시하고 수치예제를 통해 타당성을 증명한 후, 이를 공학문제에 적용한다.

In this paper, it is proposed to constant power control of solar power system. The solar power system has disadvantage that low power density and variable power output. Proposed strategy is controled by DC/DC converter using phase shift PWM and I-PD type control applying type 1 digital system. The validity of proposed control strategy is verified from simulations results using PSIM.

Purpose: This study aims to identify the critical input variables that significantly affect the optimization of repair parts in complex weapon systems. Methods: A total of 75 input variables are initially considered and then reduced to 28 based on expert review and system requirements. An orthogonal array-based design of experiments is applied to perform sensitivity analysis. Results: The analysis reveals that several variables, including Annual Operating Hours, System Life Cycle, MTBF, Unit Price, Weight, Procurement Lead Time, Unit Repair Cost, and Repair Time, have a strong influence on the total life cycle cost. Conclusion: By applying orthogonal arrays to sensitivity analysis, the study identifies key variables that significantly affect optimization outcomes. These variables should be managed with precision through enhanced user training and system guidance. Meanwhile, less sensitive inputs may be simplified, improving efficiency and reducing the potential for data entry errors in logistics support systems.

This study were conducted to evaluate the mass balance of phosphorus and nitrogen with cropping in experimental pot where the mass of input concentration, and fertilization were variable. Four treatments include CSWNF, TWCF, SWNF. And these cases were compared to the control case of tap water irrigation with conventional fertilization (CONTROL). Fertilization were following conventional fertilization , N : P : K = 11kg : 7kg : 8kg. Input water loading was CSWNF (N:48.7g ,P:3.6g), TWCF(7.8g, 0.6g), SWHF(38.8g ,2.9g), TWNF(38.8g, 2.9g ) and CONTROL(0g ,0g) The result is nitrogen decrease rate; TWCF(19.2%), SWHF(14.9%), CSWNF(9.2%) and SWCF(5.6%). phosphorous decrease rate ; TWCF (10%), SWHF(3.7%), SWCF(0.9%) and CSWNF(0.3%).

In this paper, the fuzzy control with variable input gain is applied to maintain the consistent tension in the process of taking up and releasing texture. In the process of discharging web on one side rolling it on another, the take-up drum gets smaller on the release drum side as it gets bigger on the rolling side, thus it is necessary to change the balance of velocity between the sides. In order to solve the problem a tension controller is necessary. The PI control method has been employed to maintain the consistent tension, but the PI control method produces a problem which requires an experienced worker with the traits of the machine, in order to perform the fine adjustments according to the environment of the process. For solving the above problem, we apply fuzzy control to the tension system, in order to produce a uniform roll. For the performance test, the fuzzy controller does not need to revise the parameters. Therefore the fuzzy controller exhibits an excellent additivity for the tension system where the system is changed with time.

In this paper, a new phase detector which can detect phase difference of variable input frequency and represent as a DC voltage is designed. The proposed phase detector has detection range from $-180^{\circ}$ to $180^{\circ}$. It is implemented by digital electronic circuit. It operates from 125 kHz to 4 MHz frequency of input signal and it's maximum phase error is $360/256^{\circ}$.

지난 10여 년간 딥러닝(Deep Learning)은 다양한 기계학습 알고리즘 중에서 많은 주목을 받아 왔다. 특히 이미지를 인식하고 분류하는데 효과적인 알고리즘으로 알려져 있는 합성곱 신경망(Convolutional Neural Network, CNN)은 여러 분야의 분류 및 예측 문제에 널리 응용되고 있다. 본 연구에서는 기계학습 연구에서 가장 어려운 예측 문제 중 하나인 주식시장 예측에 합성곱 신경망을 적용하고자 한다. 구체적으로 본 연구에서는 그래프를 입력값으로 사용하여 주식시장의 방향(상승 또는 하락)을 예측하는 이진분류기로써 합성곱 신경망을 적용하였다. 이는 그래프를 보고 주가지수가 오를 것인지 내릴 것인지에 대해 경향을 예측하는 이른바 기술적 분석가를 모방하는 기계학습 알고리즘을 개발하는 과제라 할 수 있다. 본 연구는 크게 다음의 네 단계로 수행된다. 첫 번째 단계에서는 데이터 세트를 5일 단위로 나눈다. 두 번째 단계에서는 5일 단위로 나눈 데이터에 대하여 그래프를 만든다. 세 번째 단계에서는 이전 단계에서 생성된 그래프를 사용하여 학습용과 검증용 데이터 세트를 나누고 합성곱 신경망 분류기를 학습시킨다. 네 번째 단계에서는 검증용 데이터 세트를 사용하여 다른 분류 모형들과 성과를 비교한다. 제안한 모델의 유효성을 검증하기 위해 2009년 1월부터 2017년 2월까지의 약 8년간의 KOSPI200 데이터 2,026건의 실험 데이터를 사용하였다. 실험 데이터 세트는 CCI, 모멘텀, ROC 등 한국 주식시장에서 사용하는 대표적인 기술지표 12개로 구성되었다. 결과적으로 실험 데이터 세트에 합성곱 신경망 알고리즘을 적용하였을 때 로지스틱회귀모형, 단일계층신경망, SVM과 비교하여 제안모형인 CNN이 통계적으로 유의한 수준의 예측 정확도를 나타냈다.

Over the past decade, deep learning has been in spotlight among various machine learning algorithms. In particular, CNN(Convolutional Neural Network), which is known as the effective solution for recognizing and classifying images or voices, has been popularly applied to classification and prediction problems. In this study, we investigate the way to apply CNN in business problem solving. Specifically, this study propose to apply CNN to stock market prediction, one of the most challenging tasks in the machine learning research. As mentioned, CNN has strength in interpreting images. Thus, the model proposed in this study adopts CNN as the binary classifier that predicts stock market direction (upward or downward) by using time series graphs as its inputs. That is, our proposal is to build a machine learning algorithm that mimics an experts called 'technical analysts' who examine the graph of past price movement, and predict future financial price movements. Our proposed model named 'CNN-FG(Convolutional Neural Network using Fluctuation Graph)' consists of five steps. In the first step, it divides the dataset into the intervals of 5 days. And then, it creates time series graphs for the divided dataset in step 2. The size of the image in which the graph is drawn is $40(pixels){\times}40(pixels)$, and the graph of each independent variable was drawn using different colors. In step 3, the model converts the images into the matrices. Each image is converted into the combination of three matrices in order to express the value of the color using R(red), G(green), and B(blue) scale. In the next step, it splits the dataset of the graph images into training and validation datasets. We used 80% of the total dataset as the training dataset, and the remaining 20% as the validation dataset. And then, CNN classifiers are trained using the images of training dataset in the final step. Regarding the parameters of CNN-FG, we adopted two convolution filters ($5{\times}5{\times}6$ and $5{\times}5{\times}9$) in the convolution layer. In the pooling layer, $2{\times}2$ max pooling filter was used. The numbers of the nodes in two hidden layers were set to, respectively, 900 and 32, and the number of the nodes in the output layer was set to 2(one is for the prediction of upward trend, and the other one is for downward trend). Activation functions for the convolution layer and the hidden layer were set to ReLU(Rectified Linear Unit), and one for the output layer set to Softmax function. To validate our model - CNN-FG, we applied it to the prediction of KOSPI200 for 2,026 days in eight years (from 2009 to 2016). To match the proportions of the two groups in the independent variable (i.e. tomorrow's stock market movement), we selected 1,950 samples by applying random sampling. Finally, we built the training dataset using 80% of the total dataset (1,560 samples), and the validation dataset using 20% (390 samples). The dependent variables of the experimental dataset included twelve technical indicators popularly been used in the previous studies. They include Stochastic %K, Stochastic %D, Momentum, ROC(rate of change), LW %R(Larry William's %R), A/D oscillator(accumulation/distribution oscillator), OSCP(price oscillator), CCI(commodity channel index), and so on. To confirm the superiority of CNN-FG, we compared its prediction accuracy with the ones of other classification models. Experimental results showed that CNN-FG outperforms LOGIT(logistic regression), ANN(artificial neural network), and SVM(support vector machine) with the statistical significance. These empirical results imply that converting time series business data into graphs and building CNN-based classification models using these graphs can be effective from the perspective of prediction accuracy. Thus, this paper sheds a light on how to apply deep learning techniques to the domain of business problem solving.