• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.9C
• /
• pp.887-898
• /
• 2005

In this paper, we introduce a probabilistic model based on Bayesian networks BNs) for recognizing human fatigue. First of all, we measured face feature information such as eyelid movement, gaze, head movement, and facial expression by IR illumination. But, an individual face feature information does not provide enough information to determine human fatigue. Therefore in this paper, a Bayesian network model was constructed to fuse as many as possible fatigue cause parameters and face feature information for probabilistic inferring human fatigue. The MSBNX simulation result ending a 0.95 BN fatigue index threshold. As a result of the experiment, when comparisons are inferred BN fatigue index and the TOVA response time, there is a mutual correlation and from this information we can conclude that this method is very effective at recognizing a human fatigue.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.36 no.3B
• /
• pp.248-253
• /
• 2011

Collision avoidance algorithm of vessels have been studied to avoid collision and grounding of a vessel due to human error. In this paper, We propose a collision avoidance algorithm using bayesian estimation theory for safety sailing and reduced risk of collision accident. We calculate collision risk for efficient collision avoidance using bayesian algorithm and determined the safest and most effective collision risk is predicted by using re-planned with re-evaluated collision risk in the future(t=t'). Others ship position is assumed to be informed from AIS. Experimental results show that we estimate the safest and most effective collision risk.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.06b
• /
• pp.112-114
• /
• 2006

최근 영상 인식 정보를 서비스 로봇 도메인에서 사용하기 위한 연구와 함께 전통적인 영상 인식 방법의 성능을 높이기 위한 연구가 진행되고 있다. 기존의 방법들은 기하학적 모델을 기반으로 예측 가능한 환경에서 상황을 인식하였기에 이를 실내 환경과 같은 동적인 환경에 적용하는 것은 정확도나 인식의 효율 면에서 한계를 갖는다. 이에 지식 기반 접근 방법을 통해 정확도를 항상 시키거나 계산 비용을 감소시킴으로써 영상 인식성능을 높이기 위한 다양한 연구가 있어 왔다. 본 논문에서는 서비스 로봇이 물체를 탐색할 때, 대상 물체가 다른 물체에 의해 가려짐으로써 발생하는 불확실한 상황을 해결하기 위한 방법을 제안한다. 제안하는 방법은 발견된 물체를 컨텍스트 정보로 사용하여 대상 물체의 존재 여부를 추론하며, 이를 위해 신뢰도를 모델링할 수 있는 확률적 모델인 베이지안 네트워크와 도메인 지식을 모델링 할 수 있는 온톨로지를 함께 사용한다. 효과적인 모델링을 위해 본 논문에서는 기본적인 물체 관계를 모듈화 하여 설계하기 위한 베이지안 네트워크 구조와 확률 값 선정 방법. 이들을 온톨로지를 기반으로 주어진 상창에 따라 결합하는 방법을 제안한다. 이는 물체 관계를 모델링할 때 발생하는 중복 설계를 감소시켜주고 유지 및 보수를 용이하게 한다. 설계된 추론 모듈은 실험 결과 5가지 장소에서 높은 정확도를 보여주었다.

• Journal of KIISE:Computing Practices and Letters
• /
• v.11 no.1
• /
• pp.81-91
• /
• 2005

Today, mobile phones have become an essential item for man-to-man communication. As more people use mobile phones, various services based on mobile phone networks and high-end devices have been developed. In addition, with the growth of the concept of ubiquitous computing, there are many ongoing studies on novel and useful services in smartphone. In this paper, for personalized service in smartphone we propose an intelligent agent that uses user modeling based on bayesian network and rule based service selection mechanism. It infers the user's status such as his current affect, how he is busy, and how someone is familiar with him from personal information and communication history using bayesian network and Provides appropriate services on the basis of the inferred information. We apply it to some realistic situation to confirm the usefulness our proposed agent.

• Journal of Korea Society of Industrial Information Systems
• /
• v.25 no.4
• /
• pp.9-16
• /
• 2020

Pronunciation recognition systems that use only video information and ignore voice information can be applied to various customized services. In this paper, we develop a system that applies a Bayesian classifier to distinguish Korean vowels via lip shapes in images. We extract feature vectors from the lip shapes of facial images and apply them to the designed machine learning model. Our experiments show that the system's recognition rate is 94% for the pronunciation of 'A', and the system's average recognition rate is approximately 84%, which is higher than that of the CNN tested for comparison. Our results show that our Bayesian classification method with feature values from lip region landmarks is efficient on a small training set. Therefore, it can be used for application development on limited hardware such as mobile devices.

• The KIPS Transactions:PartB
• /
• v.9B no.5
• /
• pp.625-634
• /
• 2002

In this paper, we conducted various experiments with Bayesian networks in order to analyze clinical data of infertility patients. With these experiments, we tried to find out inter-dependencies among important factors playing the key role in clinical pregnancy, and to compare 3 different kinds of Bayesian network classifiers (including NBN, BAN, GBN) in terms of classification performance. As a result of experiments, we found the fact that the most important features playing the key role in clinical pregnancy (Clin) are indication (IND), stimulation, age of female partner (FA), number of ova (ICT), and use of Wallace (ETM), and then discovered inter-dependencies among these features. And we made sure that BAN and GBN, which are more general Bayesian network classifiers permitting inter-dependencies among features, show higher performance than NBN. By comparing Bayesian classifiers based on probabilistic representation and reasoning with other classifiers such as decision trees and k-nearest neighbor methods, we found that the former show higher performance than the latter due to inherent characteristics of clinical domain. finally, we suggested a feature reduction method in which all features except only some ones within Markov blanket of the class node are removed, and investigated by experiments whether such feature reduction can increase the performance of Bayesian classifiers.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.06b
• /
• pp.148-150
• /
• 2006

최근 모바일 디바이스의 기능이 다양해지면서 현대인에게 없어서는 안 될 필수품이 되었다. 모바일 디바이스의 사용영역이 널어지면서 늘어나는 개인 정보의 활용에 대한 관심이 집중되고 있다. 본 논문에서는 모바일 디바이스에서 사용자의 행동 패턴 분석 및 요약을 위한 지능형 에이전트를 제안한다 사용자의 다양한 행동 및 상태 패턴 분석을 위해 협력적 모듈 베이지안 네트워크를 사용한다. 협력적 모들 베이지안 네트워크는 비슷한 유형의 패턴끼리 모듈로 설계해 상호 협력적으로 작동하여 사용자의 특이성을 추론한다. 사용자에 기억에 남을 만한 특이성를 선택하기 위해 Noisy-OR gate를 적응하여 계산한 특이성간의 연결 강도와 특이성의 우선순위를 바탕으로 사용자의 하루 동안의 행동을 요약하여 구성한다. 추론을 위한 프로토타입을 작성하고 시나리오를 바탕으로 제안한 방법의 유용성을 보인다.

• Journal of the Korean Data and Information Science Society
• /
• v.12 no.2
• /
• pp.11-25
• /
• 2001

There are several algorithms for classification in modeling relations, patterns, and rules which exist in data. We learn to classify objects on the basis of instances presented to us, not by being given a set of classification rules. The Bayesian learning uses the probability distribution to express our knowledge about unknown parameters and update our knowledge by the law of probability as the evidence gathered from data. Also, the neural network models are designed for predicting an unknown category or quantity on the basis of known attributes by training. In this paper, we compare the misclassification error rates of Bayesian Neural Network method with those of other classification algorithms, CHAID, CART, and QUBST using several data sets.

• The KIPS Transactions:PartC
• /
• v.13C no.6 s.109
• /
• pp.733-740
• /
• 2006

Spam Message has been Causing widespread damages on the Internet. One source of the problems is rooted from an anonymously posted message in the bulletin board in Internet communities. This type of the Spam messages tries to advertise products, to harm other's reputation, to deliver religious messages and so on. In this paper we present the Spam message filtering using the Bayesian approach. In order to increase usefulness of the Spam filter in the bulletin board in Internet communities, we made the Spam filter which can divide the Spam message into six categories such as advertisement, pornography, abuse, religion and other. The test conducted against messages posted on the popular web sites.

• Bioinformatics and Biosystems
• /
• v.1 no.1
• /
• pp.38-45
• /
• 2006

We present a client-server system architecture for inferring genetic interaction networks based on Bayesian networks. It is typical to take tens of hours when genome-wide large-scale genetic interaction networks are inferred in the form of Bayesian networks. To deal with this situation, batch-style distributed system architectures are preferable to interactive standalone architectures. Thus, we have implemented a loosely coupled client-server system for network inference and user interface. The network inference consists of two stages. Firstly, the proposed method divides a whole gene set into overlapped modules, based on biological annotations and expression data together. Secondly, it infers Bayesian networks for each module, and integrates the learned subnetworks to a global network through common genes across the modules.