Earticle

최근 커넥티드 카는 자동차 업계 뿐 아니라 통신업체 등 다양한 산업분야에서 큰 주목을 받고 있다. 자동차도 하나 의 디바이스로 인식되고 다양한 기기와 연동되면서 상호 작용하는 시스템으로 발전하고 있고 새로운 형태의 사업기회 가 생겨나고 있다. 하지만 차량용 기술 발전에 따라 사고의 위험성 또한 대두되고 있으며 운전자의 사고를 예방하기 위 한 다양한 기술들이 여러 기관에서 연구되고 있다. 본 연구에서는 운전자의 사고를 예방하기 위한 운전부하에 대하여 연구하고, 선행 된 연구에 바탕을 두고 운전자의 행동을 판단하여 운전부하를 측정하는 개발하였다. 개발된 알고리즘은 도로 주행 실험을 통하여 다양한 도로상황, 운전자의 상태에 따른 운전행동과 운전부하에서 유효하게 작동함을 확인할 수 있었다.

In recent days, a Connected Car has caught an attention of the motor companies and various industrial institutes such as communication company. An automobile is regarded as a device and has been developed as an interactive system because the system is connected with various device. This drives a new business item, too. As a new automatic car technology is emerging, a new type of car accident is appeared, too. So, many researches for preventing car accident comes from the driver’s are carried out in many car related institutes. In this paper, we study a driver’s driving workload and develop an algorithm that measures the driver's driving behavior. We can see that the developed algorithm runs well by the experiment of road test. This results affects various road condition, driver’s driving behavior and load that reflects the driver’s status.

This paper introduces the algorithm developed to measure the local temperature of phases in the non-isothermal two-phase flow by using micro-thermocouple, which was verified using an optical chopper and a laser. The micro-thermocouple, whose outer diameter of which is 12.7 $\mu$m, consists of a couple of alumel-chromel wire (K-type). The hot-junction, a part of the sensor used to for measuring temperature, is fabricated by hot-wire shapes. The response time of the micro-thermocouple, which was measured by dynamic calibration method, was several milliseconds in flow condition. An algorithm to calculate the temperature of each, which is based on the response time of micro-thermocouple and the exponential regression method, was developed. And the developed algorithm was verified through experiments using an optical chopper and a laser.

유사성행렬은 MDS 분석 등 다양한 다차원적 분석의 기초자료로 활용된다. 유사성행렬을 얻는 방법으로 항목의 수가 상대적으로 적을 때는 쌍대비교법이 사용될 수 있으나 항목의 수가 많을 경우 분류과제가 권장된다. 본 연구에서는 분류과제의 한 형태로서 지정분류과제의 특성과 제약을 두 개의 연구를 통해 알아보았다. 실험 1에서는 분류 순서가 MDS의 차원구조에 영향을 주는지를 알아보기 위해 분류 순서를 달리 한 두 종류의 지시문에서 얻은 MDS 차원구조를 비교하였는데 MDS 평면상에서 항목들의 상대적 위치는 서로 다르지 않았다. 목록 구성을 달리해서 목록구성의 영향을 분석한 실험 2의 결과는 각각의 목록구성에 합당한 차원구조를 산출하였다. 두 연구의 결과는 지정분류과제가 분류순서에 영향을 받지 않으면서 안정된 차원구조를 보여주며, 항목집합에 내재된 차원구조를 잘 드러낸다는 것이다. 이 결과를 근거로 지정분류과제의 몇 가지 특성과 제약에 관하여 논의하였다.

A similarity matrix provides basic data for multidimensional scaling analysis such as MDS and HCA. A similarity matrix can be obtained through paired comparison methods or sorting tasks. A sorting task is preferred when the number of item is large. We investigated the traits and constraints of a bounded sorting task, which is a subcategory of sorting task, in two studies. In Exp. 1, we compared two MDS planes from different sorting orders, but found no difference in the axis values. In Exp. 2, we compared two MDS planes from the different item constructions, and found a difference in the interpretation of axis meaning inherent in item construction. The results of the two studies showed that bounded sorting tasks produce stable dimensions free from sorting order, and sensible dimensions sensitive to the item construction. On the basis of these findings some traits and constraints were discussed.

현재까지, 다양한 위험요소를 내포하고 있는 도로 곡선부의 위험도를 교통안전 측면에서 합리적이고 정량적으로 판단하는 기준이 부재한 실정이다. 이에 본 연구에서는, 지방부 2차로 도로를 대상으로, 곡선부 사고요인을 기하구조, 시선유도시설, 속도 및 운전자 사고요인으로 분류하여, 곡선부 사고와의 상관관계 분석을 실시하고, 곡선부 사고와 밀접한 관련이 있는 주요 곡선부 사고요인을 파악하여, 곡선부 위험도 판정척도를 제시하였으며, 곡선부 위험도 판정척도를 이용하여, 곡선부 위험도 판정 알고리즘을 개발하였다. 또한, 실무활용에 용이하게 하기 위하여 Worksheet을 제시하였다. 본 연구에서 개발된 곡선부 위험도 판정척도 및 알고리즘을 이용하여, 도로 곡선부 설계시 곡선부 사고요인을 사전에 파악할 수 있으며, 기존도로 개선사업 수행 시 사업우선순위 선정에도 활용될 수 있을 것으로 판단된다.

At present, there is a no rational MOE for evaluating hazard level at curve sections. Therefore, this study focus on developing the MOE and algorithm for hazard level at curve sections. The scopes of this study limited to rural two-way roads. Actual data used is accident, geometric features, safety facilities of the selected sites at curve sections. In order to develop MOE for hazard level at curve sections, accident contributing factors were classified by road geometry, visual guidance facility, speed and driver factor. A relationship between the four factors mentioned and accidents was conducted. And, the MOE for hazard level at curve sections was derived from the previous relationship analysis, and the algorithm for hazard level was developed. Finally, worksheets were suggested based on the MOE and algorithm for road designers. These developed MOE and algorithm can be used to reduce serious accident contributing factors when designing roads and also, these will be used to determine an order of priority when reconstructing roads.

This work deals with finding a suitable position correction algorithm of industrial robot based on measuring particular points, which calculates two dimensional correction quantities and the must allow visually acceptable door-chassis assembly task. Three optimizing algorithms corresponding to three differ- ent error based performance indices are compared and selected to the best one, in terms of the predefined total uniformity, line uniformity and computational time. The selected algorithm(Total Error Minimization) is implemented for a simple door-chassis model to show its effectiveness.

통신 모바일 단말기에서 어휘 인식 시스템은 부정확한 어휘로부터 음소 특징을 추출하기 때문에 음소를 인식하지 못하거나 유사한 음소 오인식 오류로 인한 낮은 인식률의 문제점을 가진다. 이러한 문제를 해결하기 위해서, 본 논문에서는 입력 음소는 음소 유사율 처리를 통해 음소 사이의 거리를 측정하여 수치로 나타내고, 신뢰도 측정을 통하여 인식되어진 결과를 확인하는 시스템을 제안하였다. 이로 인해 부정확한 어휘 제공으로 인한 오인식 오류를 최소화하였으며 음소 유사율과 신뢰도를 이용하여 오류 보정율을 구하였다. 기존 방법인 에러 패턴 학습을 이용한 시스템과 의미기반을 이용한 시스템의 성능 평가 결과 2.7%의 인식 향상율을 보였다.

In the communication mobile terminal, Vocabulary recognition system has low recognition rates, because this problems are due to phoneme feature extract from inaccurate vocabulary. Therefore they are not recognize the phoneme and similar phoneme misunderstanding error. To solve this problem, this paper propose the system model, which based on the two step process. First, input phoneme is represent by number which measure the distance of phonemes through phoneme likelihood process. next step is recognize the result through the reliability measure. By this process, we minimize the phoneme misunderstanding error caused by inaccurate vocabulary and perform error correction rate for error provrd vocabulary using phoneme likelihood and reliability. System performance comparison as a result of recognition improve represent 2.7% by method using error pattern learning and semantic pattern.

Most existing calculations of similarities suffer from data sparsity and poor prediction quality problems. For this issue, we proposed a similarity measurement algorithm based on entropy. The entropy is computed by the difference of two users’ ratings, and we also consider the size of their common rated items, the size is bigger, the weight of their similarity is higher. Experiments show that the algorithm effectively solves the problem of the inaccuracy of similarities in data sparsity or small size neighborhood environments, and outperforms other state-of-the-art CF algorithms and it is more robust against data sparsity.

Heterogeneous data stream clustering is an important issue in data stream mining, for the accuracy of the existing heterogeneous clustering algorithm is not high, and don’t have a common distance measure method, a heterogeneous data stream clustering algorithm based on the density with mixed distance measure method is proposed. HDSDen algorithm adopts an online/offline two-stage processing framework. According to the situation of dominant property, the online stage use corresponding distance measure method to define the core points among the arriving points, the purpose of the different distance calculation method is to reduce the influence of the non-dominant property on the whole clustering accuracy. All the density-reachable points form a cluster in the offline stage, and put all the not-clustered points into the reservoir, and the number of the reservoir exceeds the threshold value, we will re-cluster the points to improve the accuracy of clustering. Experiments on real data sets show that the algorithm can achieve better clustering results, and give the clustering results at any time, which can deal with the heterogeneous data stream efficiently.

Rough K-means algorithm has shown that it can provides a reasonable set of lower and upper bounds for a given dataset. With the conceptions of the lower and upper approximate sets, rough k-means clustering and its emerging derivatives become valid algorithms in vague information clustering. However, the most available algorithms ignore the difference of the distances between data objects and cluster centers when computing new mean for each cluster. To solve this issue, an improved algorithm of rough k-means clustering based on variable weighted distance measure is presented in this article. Comparative experimental results of real world data from UCI demonstrate the validity of the proposed algorithm.

본 논문에서는 거리계산 보정 알고리즘을 이용하여 거리 인식 측정을 LED 디밍제어를 통해 구현하였다. 원거 리에서 측정되는 RSSI 값의 오차를 줄이기 위해 RSSI 평균 필터링과 Feedback 필터링에 대한 계산 값을 산출 적용 시켰다. 평균 필터링을 통한 RSSI 값과 Feedback, 필터링의 계수 값을 0.5로 설정하여 측정한 RSSI 값이 일반적인 측정에 비해 최대 –61dBm에서 최소 –52.5dBm으로 약 –2dBm에서 –6dBm 정도로 불규칙하고 높은 값이 다소 감 소하는 것을 확인하였다. 정확도 향상을 위한 거리계산 보정 알고리즘을 응용하였으며 이를 통하여 거리가 증가함에 따라 오차의 범위가 감소하는 것을 확인하였다. 최종적으로 RSSI 측정결과 필터링을 이용해 불안정한 신호를 보정하였 으며, 오차 범위를 줄이기 위해 거리계산 보정 알고리즘을 적용 시행하였다. 또한 거리판별과 신호의 안정도를 표시하 기 위해 LED로 RGB 색상을 구현하였다.

In this paper, Distance recognition measurement using distance calculation correction algorithm, was realization through LED dimming control. The calculation values for the RSSI average filtering and the RSSI feedback filtering were calculated and applied to reduce the error of the RSSI value measured from a long distance. It was confirmed that the RSSI values through the average filtering and the RSSI values measured by setting the coefficient value of the feedback filtering to 0.5 were ranged from –61 dBm to - 52.5 dBm, which shows irregular and high values decrease slightly as much as about –2 dBm to -6 dBm as compared to general measurements. A distance calculation correction algorithm to improve the accuracy was applied, which confirmed that as the distance increases, the range of errors decreases. In conclusion, unstable signals were corrected using the RSSI measurement result filtering, and the distance calculation correction algorithm was applied and performed to reduce the range of errors. In addition, RGB colors were implemented by LED to indicate the distance determination and the signal stability.

We propose an efficient method to measure the insurance risk of causality insurance companies by using the CreditRisk+ methodology. This method is superior to previous methods in several aspects. Its computation speed is very fast and the input data form is simple. It is able to aggregate both credit risk and insurance risk, so the insurance company can manage the risk in combined manner. In this paper, we propose a mathematical method to obtain the aggregate loss distribution of portfolios having correlation among products or business lines as a general case, and then suggest its implementation algorithm. Finally we apply this method to the real data from Korea Insurance Development Institute (KIDI) and discuss its availability to real applications.

1. Introduction of Camera Detection system Camera Detection system is an equipment that can detect realtime traffic information by image processing techniques. This information can be used to analyze and control road traffic flow. It is also used as a method to detect and control traffic flow for ITS(Intelligent Transportation System). Traffic information includes speed, head way, traffic flow, occupation time and length of queue. There are many detection systems for traffic data. But video detection system can detect multiple lanes with only one camera and collect various traffic information. So it is thought to be the most efficient method of all detection system. Though the...

In this paper, an algorithm to reconstruct two orthogonal images into a three-dimensional image is developed in order to measure the bubble size and volume in a two-phase boiling flow. The central-active contour model originally proposed by P. $Szczypi\'{n}ski$ and P. Strumillo is modified to reduce the dependence on the initial reference point and to increase the contour stability. The modified model is then applied to the algorithm to extract the object boundary. This improved central contour model could be applied to obscure objects using a variable threshold value. The extracted boundaries from each image are merged into a three-dimensional image through the developed algorithm. It is shown that the object reconstructed using the developed algorithm is very similar or identical to the real object. Various values such as volume and surface area are calculated for the reconstructed images and the developed algorithm is qualitatively verified using real images from rubber clay experiments and quantitatively verified by simulation using imaginary images. Finally, the developed algorithm is applied to measure the size and volume of vapor bubbles condensing in a subcooled boiling flow.

This paper describes a complexity measure algorithm based on nonlinear dynamics(chaos theory). In order to quantify complexity or regularity of biomedical signal, this paper proposed fractal dimension-1 and fractal dimension-2 algorithm with digital filter. Approximate entropy algorithm which measure a system regularity are also compared. In this paper investigate what we quantify of biomedical signal. These quantified complexity measure may be a useful information about human physiology.

Line constants are composed by impedances and susceptances. We don't need all line constants on electric railway system. Depending on where they are applied, The algorithm to get line constants Is different. We suggest two algorithm to measure them.

Nowadays, to prevent electrical accidents in Korea, inspectors directly performed checking general electrical facilities as a cycle from every one to three years. It is difficult to presuppose an omen because intact conditions of electrical equipments are not kept at the time of inspection. In this paper, in order to ensure effectiveness of an online basis electric inspection, we developed an electrical safety IoT system using LoRa communication technology to enable monitoring mainly electrical safety components such as overcurrent, overvoltage, resistive leakage current, and power. Then we proposed a method for verifying performances of the prosed electrical safety IoT system. Resistive leakage currents are calculated by using difference of phase between voltages and currents. We verified that average errors are 0.97%, which reference goal is ${\pm}5%$ for a device, through reliability test according to conditions. Results of this research can be used as basic study materials to develop technologies for measuring three phase leakage current and for implementing public electrical safety. platform.

현대 정보통신기술(ICT)은 인공지능을 등장시키고 발전시킴으로써 인간 사회의 패러다임에 큰 변화를 가져오고 있다. 인공지능 발전의 가장 핵심적인 기술은 빅데이터와 알고리즘이다. 빅데이터의 알고리즘 시스템은 정교한 프로세스를 사용한다. 또 이러한 프로세스에는 데이터의 입력이 필요하다. 빅데이터와 알고리즘 기술을 가장 많이 활용하는 주체는 기업이며, 정당 등 정치적 목적을 가진 분야에서도 활발하게 이용되고 있다. 만약 알고리즘 시스템의 설계자가 데이터의 입력과 조작 등에 있어 자신의 특정한 성향이나 세계관만을 따른다면, 인공지능 알고리즘의 자동화된 의사결정과 그 생성물은 편향성, 차별성, 불공정성 등을 나타내게 된다. 이때 알고리즘은 이념의 생성물로서 필연적으로 정치적인 속성을 내포하게 된다. 알고리즘에 의한 자동화된 의사결정이 제공하는 결과가 전적으로 중립적이라고 볼 수는 없게 되는 것이다. 그 결과가 객관성과 공정성을 담아내지 못함에도 불구하고, 그 결과가 사회에 미치는 영향은 정치․경제․문화 등의 모든 분야를 망라하며, 이용자들의 선호와 판단 그리고 행동까지도 변화하게 만든다. 오늘날 알고리즘은 국가와 사회의 정치적, 사회적, 경제적, 문화적 상호작용과 의사결정에 중요한 기능과 역할을 담당하면서 계속해서 영향력을 높여 나가고 있다. 또한, 그 자체가 하나의 거대한 정치권력이 되어가는 모습마저도 나타낸다. 따라서 편향적인 데이터와 이를 통해 학습한 인공지능 알고리즘이 만들어내는 부정적인 결과인 차별, 편견, 배제, 불공정성 등을 효과적으로 통제하는 방안을 모색하는 사회적 논의와 법적 대응책이 필요하다. 이에 EU의 일반정보보호규정(General Data Protection Regulation: GDPR)의 도입과 발효(2018년 5월)는 알고리즘의 투명성과 책무성에 대한 규범적 쟁점과 논의를 확대하는 계기를 마련해 주었다. GDPR은 정보 주체가 프로파일링에 대해 반대할 권리와 자동화된 의사결정의 대상이 되지 않을 권리가 있음을 규정하고 있다. 또한, 정보 주체가 그들에 관하여 수집된 정보에 접근할 권리가 있으며, 수집된 개인정보에 관해 통지를 받을 권리가 있음을 규정하고 있다. 이러한 규정들과 이의 해석을 통해서, 알고리즘에 기반한 자동화된 의사결정에 대한 설명을 요구할 권리(Right to explanation)가 도출될 수 있다. 설명을 요구할 권리는 이용자에게 프로파일링 기반의 알고리즘 결정에 관해 설명을 요구할 기회를 부여함과 동시에, 알고리즘 설계과정에서 인간적 개입의 중요성을 강조한 점에서 큰 의미가 있다. 또한, 설명을 요구할 권리는 해당 기업의 영업 비밀에 해당하는 알고리즘 설계과정에 정보 주체가 개입할 수 있도록 강제하는 규범적 가능성을 열어줌으로써, 알고리즘의 공정성․투명성 및 차별성 이슈를 풀어나가는데 핵심적인 역할을 할 것으로 생각된다.

Modern information and communication technologies(ICTs) are bringing huge changes to the paradigms of human society by presenting and developing artificial intelligence(AI). The very core technologies in the development of AI are Big Data and algorithms. The algorithm system of Big Data uses an exquisite process, which requires the input of data. Big Data and algorithm technologies are most used by corporations and actively used in the fields with political purposes such as political parties. If the designer of an algorithm system bases data entry and manipulation only on his or her certain tendencies or world views, automated decision-making and its outcomes in the AI algorithm will display a bias, differentiation, and unfairness. In such a case, the algorithm is a product of ideology and necessarily has political attributes. There is no saying that outcomes by the automated decision-making process based on the algorithm are totally neutral. Even though such outcomes have no objectivity and fairness, their influences will encompass all the sectors of the society including politics, economy, and culture and change even the preferences, judgments and actions of users. Today, algorithms perform important functions and roles in political, social, economic, and cultural interactions and decision-making processes at the national and social level, continuing to increase their influences. They are even becoming an entity of huge political power themselves. There is thus a need for social discussions and legal measures to explore plans of effectively controlling discrimination, prejudice, exclusion, and unfairness that are negative outcomes of biased data and AI algorithms whose learning is based on biased data. EU-General Data Protection Regulation(GDPR) were introduced and came into effect in May, 2018, creating a chance to expand normative issues and discussions about the transparency and accountability of algorithms. EU-GDPR stipulate that the subjects of information have rights to refuse their profiling and not to become subjected to automated decision-making processes. EU-GDPR also state that they have rights to access to the information collected regarding them and notices about their personal information collected. The interpretations of these regulations and objections to them can result in a right to explanations about automated decision-making processes based on an algorithm. This right to explanations holds huge significance in that it grants an opportunity for users to request explanations about profiling-based algorithm decisions and puts an emphasis on the importance of human interventions in the algorithm design process. This right to explanations also opens up the normative possibilities of enforcing an intervention into the algorithm design process, which is part of a corporation's confidential business information, by the information subject, thus playing an essential role in addressing the fairness, transparency, and differentiation issues of algorithms.

깊이 추정은 로봇 비전, 3차원 영상, 모션 제어를 위해 사용되는 매우 중요한 인자이다. 깊이 추정은 렌즈와 물체 사이의 거리를 변화시켜가면서 취득된 일련의 영상에서 계산된 초점 정보에 기반을 둔다. 본 논문에서는 다양한 초점정보를 이용한 패턴간의 상대적 깊이 추정 알고리즘을 제안한다. 제안된 방법은 거리별로 취득된 영상의 초점값 정보를 이용하여 구현하였으며, 깊이는 두 패턴의 상대적 거리를 고려함으로써 추정하였다. 다양한 영상정보를 이용하여 깊이 추정을 수행한 결과 효과적인 추정이 가능함을 알 수 있었다.

Depth estimation is an essential factor for robot vision, 3D scene modeling, and motion control. The depth estimation method is based on focusing values calculated in a series of images by a single camera at different distance between lens and object. In this paper, we proposed a relative depth estimation method using focus measure. The proposed method is implemented by focus value calculated for each image obtained at different lens position and then depth is finally estimated by considering relative distance of two patterns. We performed various experiments on the effective focus measures for depth estimation by using various patterns and their usefulness.

현재 많은 조직에서 위험 분석을 통해 현재 자신들의 보안상의 문제점을 파악하고 그에 따른 대책을 적용하고 있다. 기존의 보안 대책을 평가하고 새로운 보안 대책을 적용하는 데 않은 어려움이 따르므로 본 연구에서는 Simple Genetic Algorithm을 이용하여 현재 조직의 상황에 적절한 보안 대책을 제시할 수 있는 방법을 연구하고자 한다.