• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06b
• /
• pp.510-512
• /
• 2012

상황인식 서비스 분야에서 불확실한 데이터를 추론하는 것은 매우 어렵고 복잡하다. 이러한 상황정보들에서 얻어지는 데이터는 불확실성을 내포하고 있어서 불확실한 추론 결과를 초래할 수 있다. 비록 불확실성 문제들을 해결하기 위해 퍼지 이론, 뉴런 네트워크, 동적 베이지안 네트워크, 은닉 마르코프 모델과 같은 여러 종류의 방법들이 제시되었지만 이러한 방법들은 가설들을 하나의 숫자에 의해 신뢰의 정도를 표시하기 때문에 많은 어려움이 있다. 본 논문에서는 사용자들이 제공받는 서비스들에 대하여 만족도를 평가한 후 수집된 데이터를 활용하여 사용자들의 상관 관계를 분석한다. 그리고 Dempster-Shafer 이론을 사용하여 사용자들로부터 측정된 믿음 값을 융합한다. 이는 불확실성 값을 낮추어 추론결과의 정확성을 높이고 증거구간을 재설정하여 사용자들에게 신뢰성 있는 적응형 서비스를 제공하게 한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10b
• /
• pp.83-87
• /
• 2006

업무 능률을 높이기 위해 일정의 입력이나 변경 및 삭제를 체계적인 관리하는 일정관리 에이전트에 대한 연구가 수행되고 있다. 사용자와의 지속적인 상호작용을 통해 사용자의 의도를 파악하며, 사용자에게 적절한 일정을 추천하거나 관련된 정보를 제공하는 서비스가 함께 요구된다. 본 논문에서는 상호주도형 대화를 통해 사용자의 의도를 능동적으로 추론하여 보다 자연스러운 대화를 제공하며, 상황인식을 통해 사용자에게 적절한 일정관리 서비스를 제공하는 지능형 일정관리 에이전트를 제안한다. 대화의 문맥 유지와 상호주도적 문제해결을 위한 반응형 스크립트를 설계하고, 동적 베이지안 네트워크로 일정과 관련된 사용자의 상태를 추론한다. 사용자에게 적절한 일정시간이나 장소를 추천하기 위해 최근 활발히 연구되고 있는 상식 DB인 ConceptNet을 도입하여 지능적인 일정관리 서비스를 제공한다. 각종 시나리오를 통해 제안하는 에이전트의 유용성을 검증하였다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.5
• /
• pp.988-995
• /
• 2009

One of the most popular services of Telematics is a shortest path finding from a starting point to a destination. In this paper, a dynamic shortest path finding system with forecasting result of traffic flow in the future was developed and various experiments to verify the performance of our system using real-time traffic information has been conducted. Traffic forecasting has been done by a prediction system using Bayesian network. It searched a dynamic shortest path, a static shortest path and an accumulated shortest path for the same starting point and destination and calculated their travel time to compare with one of its real shortest path. From the experiment, over 75%, the travel time of dynamic shortest paths is the closest to one of their real shortest paths than one of static shortest paths and accumulated shortest paths. Therefore, it is proved that finding a dynamic shortest path by applying traffic flows in the future for intermediated intersections can give more accurate traffic information and improve the quality of services of Telematics than finding a static shortest path applying by traffic flows of the starting time for intermediated intersections.

• Journal of KIISE:Software and Applications
• /
• v.32 no.11
• /
• pp.1071-1083
• /
• 2005

By estimating probability distributions of the good solutions in the current population, some researchers try to find the optimal solution more efficiently. Particularly, finite mixtures of distributions have a very useful role in dealing with complex problems. However, it is difficult to choose the number of components in the mixture models and merge superior partial solutions represented by each component. In this paper, we propose a new continuous evolutionary optimization algorithm with distribution estimation by variational Bayesian mixtures of factor analyzers. This technique can estimate the number of mixtures automatically and combine good sub-solutions by sampling new individuals with the latent variables. In a comparison with two probabilistic model-based evolutionary algorithms, the proposed scheme achieves superior performance on the traditional benchmark function optimization. We also successfully estimate the parameters of S-system for the dynamic modeling of biochemical networks.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.9C
• /
• pp.756-760
• /
• 2010

This paper is proposed to an novel facial behavior recognition system for driver's fatigue detection. Facial behavior is shown in various facial feature such as head expression, head pose, gaze, wrinkles. But it is very difficult to clearly discriminate a certain behavior by the obtained facial feature. Because, the behavior of a person is complicated and the face representing behavior is vague in providing enough information. The proposed system for facial behavior recognition first performs detection facial feature such as eye tracking, facial feature tracking, furrow detection, head orientation estimation, head motion detection and indicates the obtained feature by AU of FACS. On the basis of the obtained AU, it infers probability each state occur through Bayesian network.

• Journal of Navigation and Port Research
• /
• v.39 no.3
• /
• pp.149-156
• /
• 2015

The Dynamic Positioning System consists of 7 elements which are namely Power system, Human machine interface, DP Computer, Position Reference System(PRS), Sensors, Thruster system and DP Operator. Incidents like loss of position(LOP) on DP vessel usually occur due to errors in these 7 elements. The purpose of this study is to find out safety operation method of DP vessel through qualitative and quantitative analyze of DP LOP incidents which are submitted to IMCA every year. The 612 DP LOP incidents submitted from 2001 to 2010 were analyzed to find out the main cause of the incidents and its rate among other causes. Consequently, the highest rate of incidents involving DP elements are PRS errors. DP computer, Power system, Human error and thruster system came next. The PRS has been analyzed and a flowchart was drawn through expert brainstorming. Also, the conditional probability has been analyzed through Bayesian Networks based on this flowchart. Consequentially, the main causes of drive off incidents were DGPS, microwave radar and HPR. Also, this study identified the main causes of DGPS errors through Bayesian Networks. These causes are signal blocked, electric components failure, relative mode error, signal weak or fail.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.8
• /
• pp.1-10
• /
• 2011

Recently, cloud computing which has provided enabling convenience that users can connect from anywhere and user friendly environment that offers on-demand network access to a shared pool of configurable computing resources such as smart-phones, net-books and PDA etc, is to be watched as a service that leads the digital revolution. Now, when business practices between departments being integrated through a cooperating system such as cloud computing, data streaming between departments is getting enormous and then it is inevitably necessary to find the solution that person in charge and find data they need. In previous studies the clustering simplifies the search process, but in this paper, it applies Hash Function to remove the de-duplicates in large amount of data in business firms. Also, it applies Bayesian Network of data mining for classifying the respect data and presents handling cloud computing based data. This system features improved search performance as well as the results Compared with conventional methods and CPU, Network Bandwidth Usage in such an efficient system performance is achieved.

• Journal of KIISE:Software and Applications
• /
• v.37 no.10
• /
• pp.779-787
• /
• 2010

Since robots have been used widely recently, research about human-robot communication is in process actively. Typically, natural language processing or gesture generation have been applied to human-robot interaction. However, existing methods for communication among robot and human have their limits in performing only static communication, thus the method for more natural and realistic interaction is required. In this paper, an emotional gesture based dialogue management system is proposed for sophisticated human-robot communication. The proposed system performs communication by using the Bayesian networks and pattern matching, and generates emotional gestures of robots in real-time while the user communicates with the robot. Through emotional gestures robot can communicate the user more efficiently also realistically. We used behavior networks as the gesture generation method to deal with dialogue situations which change dynamically. Finally, we designed a usability test to confirm the usefulness of the proposed system by comparing with the existing dialogue system.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.9
• /
• pp.906-910
• /
• 2008

As the mobile environment becomes widely used, there is a growth on the concern about recognizing and sharing user context. Sharing context makes the interaction between human more plentiful as well as helps to keep a good social relationship. Recently, it has been applied to some messengers or mobile applications with sharing simple contexts, but it is still required to recognize and share more complex and diverse contexts. In this paper, we propose a context-aware messenger that collects various sensory information, recognizes representative user contexts such as emotion, stress, and activity by using dynamic Bayesian networks, and visualizes them. It includes a modular model that is effective to recognize various contexts and displays them in the form of icons. We have verified the proposed method with the scenario evaluation and usability test.

• Journal of Korean Society of Transportation
• /
• v.18 no.5
• /
• pp.83-97
• /
• 2000

The traffic management schemes through traffic signal control and information provision could be effective when the link-level data and trip-level data were used simultaneously in analysis Procedures. But, because the trip-level data. such as origin, destination and departure time, can not be obtained through the existing surveillance systems directly. It is needed to estimate it using the link-level data which can be obtained easily. Therefore the objective of this study is to develop the model to estimate O-D demand using only the link flows in highway network as a real time. The methodological approaches in this study are kalman filer, least-square method and normalized least-square method. The kalman filter is developed in the basis of the bayesian update. The normalized least-square method is developed in the basis of the least-square method and the natural constraint equation. These three models were experimented using two kinds of simulated data. The one has two abrupt changing Patterns in traffic flow rates The other is a 24 hours data that has three Peak times in a day Among these models, kalman filer has Produced more accurate and adaptive results than others. Therefore it is seemed that this model could be used in traffic demand management. control, travel time forecasting and dynamic assignment, and so forth.